Soil and terrain attributes for evaluation of leaching in a Montana farm field
by Melissa Ann Landon
A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Soils
Montana State University
© Copyright by Melissa Ann Landon (1995)
Abstract:
Degradation of water quality from agricultural sources is a critical concern for both surface and ground
waters in the United States. Nutrient (nitrate) management on the farm can reduce agricultural
non-point pollution. Practical and operational solutions to ground water contamination problems will
require consideration of contrasting environmental differences within fields. One difficulty in
designing leaching models is that when a model successfully simulates a particular profile, it may not
be representative of the overall field situation due to spatial and/or temporal variability.
The objectives of this study were to: 1) characterize the spatial variability of nitrate and bromide
leaching; 2) test the leaching prediction capabilities of several soil, soil water, and terrain attributes that
singly, or in combination, represent the spatial distribution of soils and microclimate in a farm field
under uniform application of nitrogen and bromide; and 3) determine if a topographically-derived
wetness index is correlated with field-measured soil water content. Nitrogen was applied by the farmer
in the course of normal field management and potassium bromide was uniformly applied to small areas
at 70 selected sites. Soil samples were collected at each site to a maximum depth of 180 cm every
spring and fall for three years (two cropped, one fallow).
Small spatial variability in soil water (to 25 cm, 100 cm and 150 cm depths) was measured. It is
possible that the thick, permeable, loess-derived soils capture and retain precipitation from rainfall and
melting snow without significant lateral distribution in the landscape. The observed variability in solute
concentrations was also small and less than reported in other field studies. The spatial variability of
bromide increased over time indicating that movement (leaching) differences may be expressed slowly.
Paired t-tests and correlation analyses indicated that the spatial behavior patterns of nitrogen and
bromide were similar in only two of five time periods.
Three types of leaching indices were calculated for both nitrogen and bromide. Stepwise multiple
regression analysis occasionally showed some promise, but the results were not consistent. None of the
three index types produced results more reliable or consistent than the other two. Exclusion of sites
within an ephemeral stream channel also did not produce substantially or consistently better results.
Overall, nitrogen leaching indices produced slightly stronger relationships with the environmental
factors than bromide indices.
Correlation analysis indicated that wetness indices at two scales (2 m and 10 m) were moderately
correlated to one another and to the thickness of the mollic horizon. Neither the wetness indices or
thickness of mollic horizon were correlated to either season average measured water content or average
measured water content from June of each year.
These results reiterate the difficulties that are encountered in trying to understand (predict) the
interrelationships between environment, nitrogen leaching, and potential contamination of ground
water. SOIL AND TERRAIN ATTRIBUTES FOR EVALUATION
OF LEACHING IN A MONTANA FARM FIELD
by
Melissa Ann Landon
A thesis submitted in partial fulfillment.
of the requirements for the degree
of
Master of Science
in
Soils
MONTANA STATE UNIVERSITY
Bozeman, Montana
May 1995
© COPYRIGHT
by
Melissa Ann Landon
1995
All Rights Reserved
■ /> 3 U 3
11
APPROVAL
of a thesis submitted by
Melissa Ann Landon
This thesis has been read by each member of the thesis committee and has been
found to be satisfactory regarding content, English usage, format, citations, bibliographic
style, and consistency, and is ready for submission to the College of Graduate Studies.
Chairperson, Graduate Committee
Date
Approved for the Major Department
_______ 57 /c / ^ r
HeacC Major Department
Date
Approved for the College of Graduate Studies
6 /3 o /f>
Date
Z
/
Graduate Dean
STATEMENT OF PERMISSION TO USE
In presenting this thesis in partial fulfillment of the requirements for a master’s
degree at Montana State University, I agree that the Library shall make it available to
borrowers under rules of the Library.
If I have indicated my intention to copyright this thesis by including a copyright
notice page, copying is allowable only for scholarly purposes, consistent with "fair use"
as prescribed in the U.S. Copyright Law.
Requests for permission for extended
quotation from or reproduction of this thesis in whole or in parts may be granted only
by the copyright holder.
Signature
Date
iv
ACKNOWLEDGEMENTS
I would like to acknowledge Drs. Jeff Jacobsen, Jerry Nielsen, and John
Wilson for serving on my committee and providing both the opportunity and funding
for not only my graduate research, but also attendance at the 1994 SSSA meetings in
Seattle, PC ARC/INFO and GPS training courses, and other workshops. Dr. Jon
Wraith, who also served on my committee, provided invaluable assistance with the
neutron probe, soil water data, and writing the LEACHIND.BAS program.
Additional thanks to Dr. John Wilson for his time, ideas and input, and assistance
with statistical data analysis and thesis structure.
This work was supported by USDA/CSRS Grant #91-34214-6057 under the
Agricultural Management and Water Quality section of the Water Quality Special
Grants Program. I would like to extend my appreciation to Mr. Bill Wright and
family for cooperation on this project and to AgriBasics in Belgrade, Montana for
their donation of fertilizer. Dr. Dave Tyler facilitated collection of GPS data and Mr.
Brian Autio re-surveyed sample site locations. Kirk McEachem performed the initial
soil sampling, profile descriptions, and installation of neutron access tubes. Thanks to
Damian Spangmd for terrain and soil attribute data exchange and preparation of
several figures.
V
ACKNOWLEDGEMENTS--Continued
Many people assisted in the completion of field work and processing of soil
samples. Thanks to Diana Cooksey, Denise Culver, Debi Duke, Heidi Hart, Julia
McHugh, Scott Lorbeef, Bemie Schaff, Brian Woodbury, and Eva Zelenak. Thanks
• >
to Debi Duke and Diana Cooksey for logistical support, comraderie, and patience in
_
.
answering innumerable computer questions.
A very special thanks to Denise Culver and Eva Zelenak for their friendship
and support over the last couple years. Finally, I would like to express my heartfelt
appreciation to my parents, Charles and Marilee Landon, for their enduring love and
encouragement.
(
vi
TABLE OF CONTENTS
Page
APPROVAL
........................................... .................. ............................. .. .
ii
STATEMENT OF PERMISSION TO USE ................ ............. ..............................
iii
ACKNOWLEDGEMENTS ................................
iv
TABLE OF CONTENTS
vi
. . . / .............................................................. ..
LIST OF TABLES ..............................................................
LIST OF FIG URES........................
ix
xiii
A B S T R A C T ........... ...................................................................................................... xiv
CHAPTER:
I. SITE-SPECIFIC FARMING, NITRATE LEACHING,
AND GROUND WATER Q U A L IT Y ............................................
I
Background . ........................
Nitrate Leaching and Ground Water Q u a lity .................................
Spatial Variability of Waterand Soil P ro p erties....................
Site-Specific F a rm in g .......................................................................
On-Farm, Field-Scale Information System s....................................
Field-Scale Solute Transport M o d els................................
Scope and P u rp o s e ................ ■. . ...............................................................
Objectives .....................................................................................................
I
I
4
6
8
8
14
15
vii
TABLE OF CONTENTS-Continued
Page
2. DESCRIPTION AND PREDICTION OF FIELD-SCALE
LEACHING.......................................................................................
Introduction.......................................................................................
O bjectives.............................................................................................
Description of Study A r e a ...............................................................
Materials and Methods ....................................................................
Field Management ...............................................................
1992 ............................................................................
1993 ............................................................................
1994 ................................................................................
Preparation of GPS Survey and Sample Grid .......................
Soil Sampling .......................................................................
Generation of Terrain A ttributes.............................................
Soil Water C o n te n t...............................................................
Grain Yield and Plant Uptake Measurements ...................
Computations for Leaching Characterization......................
Statistical Analysis ..................................................................
Results and Discussion .......................................................................
Leaching Behavior . . .............................................................
N itro g e n ....................................................................
Bromide .......................................................................
Nitrogen Behavior vs. BromideBehavior ...............
Conclusion ..................................................................
Environmental Controls .......................................................
Observed Spatial and TemporalV ariab ility ............
Soil W a te r..................................................................
S o lu te s .......................................................................
Multiple Regression Analysis....................................
N itro g e n ....................................................................
Bromide ....................................................................
D iscussion.................................................................
Conclusions.......................................................................................
16
16
21
22
26
27
27
27
29
29
32
33
36
37
38
43
45
45
45
51
60
63
63
65
66
70
73
77
89
94
98
3. RELATIONSHIP BETWEEN MEASURED SOIL WATER CONTENT
AND A TOPOGRAPHICALLY-DERIVEDWETNESS INDEX .
100
Introduction.........................................................................................
O bjectives............................................................................................
100
104
viii
TABLE OF CONTENTS-Continued
Page
Materials and Methods ...........................................
Study S ite ...............................................................................
Depth of Mollic Horizon
Generation of Wetness I n d ic e s ............................................
Soil Water C o n te n t................. .............................................
Statistical Analysis ......................................................... .. .
Results and Discussion ....................................................................
Wetness In d ic e s................................................................. .
Soil Water Content ’ ...............................................................
Correlation Analysis ............................................................
Conclusions............................................................................
104
104
104
105
105
106
107
107
109
Ill
114
\
4. SUMMARY.............................................................................................
117
REFERENCES C IT ED ...................................
120
APPEND ICES...............................................................................................
136
Appendix
Appendix
Appendix
Appendix
Appendix
A-Procedures for Plant Tissue and Soil Bromide Analysis .
B—LEACHIND.BAS Outline andP ro g ram ................................
C—Initial Soil Test and Terrain Values ....................................
D-Multiple Regression R e su lts.................................................
E—Wetness Index V alu es............................................................
137
139
149
157
166
ix
LIST OF TABLES
Table
Page
1. Review of leaching m o d e ls ...................................................................
10
2. Models for solute leaching in soils....................................................................
12
3. Springhill project tim eline..................................................................................
28
4. Mass balance worksheet factors ......................................................................
42
5. Summary statistics for nitrogen solute mass (mg kg'1) ....................................
45
6. Summary statistics for nitrogen critical depth (c m )..........................................
47
7. Summary statistics for nitrogen balance (% unrecoverable)............................
49
8. Summary statistics for bromide solute mass (mg kg'1) .....................................
53
9. Summary statistics for bromide balance (% unrecoverable)............................
55
10. Summary statistics for bromide critical depth (cm )..........................................
57
11. Comparison of nitrogen and bromide solute mass by season .........................
61
12. Comparison of nitrogen and bromide mass balance by seaso n ......................
62
13. Summary statistics for edaphic controls ........................
64
14. Summary statistics for topographic controls....................................................
64
15. Summary statistics for soil water controls: average total water ( c m ) ...........
65
16. Summary statistics for biotic controls: Fall 1992 barley yield
and plant uptake.............................................................................................
65
17. Differences between soil water season averages and June data
68
...................
X
LIST OF TABLES-Continued .
Table
Page
18. Correlations for total soil water between season averages
and June data ............................................................................................
69
19. Regression approaches to prediction of field-scale leaching
.........................
75
20. Abbreviations for soil water variables ............................................................
75
21. Abbreviations for soil, plant, and terrain attributes
......................................
77
22. Multiple regression results for nitrogen solute mass in Spring 1992 ...........
78
23. Multiple regression results for nitrogen balance to 60 cm in Fall 1992
. . .
82
24. Multiple regression results for nitrogen solute mass in Spring 1993 ...........
82
25. Multiple regression results for nitrogen balance in Spring 1993 ...................
84
26. Multiple regression results for nitrogen solute mass in Fall 1993 . .c...........
85
27. Multiple regression results for nitrogen critical depth in Fall 1993 ..............
85
28. Multiple regression results for nitrogen balance in Spring 1994 ...................
87
29. Multiple regression results for nitrogen critical depth in Fall 1994 ..............
88
30. Multiple regression results for change in nitrogen critical depth in
Fall 1994 ........................... ............................................... . ......................
88
31. Multiple regression results for change in bromide solute mass from
Spring 1992-Fall 1992 . . ............................................................ ..
89
32. Multiple regression results for bromide balance in Fall 1992
......................
90
33. Multiple regression results for bromide balance in Spring 1993 ...................
91
34. Multiple regression results for bromide critical depth in Fall 1993 . . . . . .
93
35. Multiple regression results for bromide solute mass in Fall 1994 ................
94
S
36. Summary statistics for wetness indices
.......................................................
107
Xl
LIST OF TABLES-Continued
Table
Page
37. Correlation analyses of wetness indices at 2 m and 10 m scales...................
109
38. Summary statistics for season average total water (SCI) ..............................
HO
39. Summary statistics for average total water in June (S C I)..............................
Ill
40. Pearson’s correlation analyses between wetness indices
and soil water variables (SCI) ....................................................................
112
41. Pearson’s correlation analyses between wetness indices
and soil water variables (SC O ).......................................................................
113
42. Correlation (Pearson’s and Spearman Rank) analyses between
wetness indices and thickness of mollic horizon . . . .........................
115
43. Initial soil test values
...................................................... '...............................
150
44. Measured and calculated terrain attributes .......................................................
153
45. Multiple regression results for nitrogen critical depth in Spring 1993 . . . .
158
46. Multiple regression results for change in nitrogen solute mass
from Fall 1992 - Spring 1993 . .................................................................
158
47. Multiple regression results for change in nitrogen critical depth
from Fall 1992 - Spring 1993 .......................................................................
159
48. Multiple regression results for change in nitrogen solute mass
from Spring 1993 - Fall 1993 ....................................................................
159
49. Multiple regression results for change in nitrogen critical depth
from Spring 1993 - Fall 1993 ........................ ...........................................
160
50. Multiple regression results for nitrogen balance in Fall 1993
....................
160
51. Multiple regression results for nitrogen solute mass in Fall 1994 ...............
161
52. Multiple regression results for bromide solute mass in Spring 1993 .........
161
53. Multiple regression results for bromide critical depth in Spring 1993 . . . .
162
X ll
LIST OF TABLES-Continued
Table
Page
54. Multiple regression results for change in bromide critical depth
from Fall 1992 - Spring 1993 ....................................................................
162
55. Multiple regression results for bromide solute mass in Fall 1993
........... ..
163
56. Multiple regression results for change in bromide solute mass
from Spring 1993 - Fall 1993 ....................................................................
163
57. Multiple regression results for change in bromide solute mass
from Fall 1993 - Spring 1994 ....................................................................
164
58. Multiple regression results for change in bromide critical depth
from Fall 1993 - Spring 1994 ...................................................................
164
59. Multiple regression results for bromide balance in Spring 1994 ...................
165
60. Multiple regression results for change in bromide critical depth
from Spring 1994 - Fall 1994 ....................................................................
165
61. Wetness index values
167
.......................................................................................
xiii
i
LIST OF FIGURES
Figure
Page
1. Location map .....................................................................................................
23
2. Elevation contour m a p .......................................................................................... 24
3. Interpolated 10 m elevation lattic e ..................
25
4. GPS collection p a tte rn .......................................................................................
30
5. Permanent sampling grid
31
........................................
6. LEACHIND.BAS flowchart
...........................................
40
7. Average nitrogen solute mass (mg kg'1) ........................................
46
8. Average nitrogen critical depth (cm) ......................................................... . .
48
9. Average nitrogen balance (% unrecoverable) ................
50
10. Average bromide solute mass (mg kg'1)
54
................................
11. Average bromide balance (% unrecoverable).....................
56
12. Average bromide critical depth ( c m ) ........... .. ...................... -.......................
58
13. Scatterplot of wetness index values at 2m and 10 m scales . . . ..................
108
14. LEACHIND.BAS p ro g ram ...............................................................................
141
xiv
ABSTRACT
Degradation of water quality from agricultural sources is a critical concern for
both surface and ground waters in the United States. Nutrient (nitrate) management on
the farm can reduce agricultural non-point pollution. Practical and operational solutions
to ground water contamination problems will require consideration of contrasting
environmental differences within fields. One difficulty in designing leaching models is
that when a model successfully simulates a particular profile, it may not be representative
of the overall field situation due to spatial and/or temporal variability.
The objectives of this study were to: I) characterize the spatial variability of
nitrate and bromide leaching; .2) test the leaching prediction capabilities of several soil,
soil water, and terrain attributes that singly, or in combination, represent the spatial
distribution of soils and microclimate in a farm field under uniform application of
nitrogen and bromide; and 3) determine if a topographically-derived wetness index is
correlated with field-measured soil water content. Nitrogen was applied by the farmer
in the course of normal field management and potassium bromide was uniformly applied
to small areas at 70 selected sites. Soil samples were collected at each site to a
maximum depth of 180 cm every spring and fall for three years (two cropped, one
fallow).
Small spatial variability in soil water (to 25 cm, 100 cm and 150 cm depths) was
measured. It is possible that the thick, permeable, loess-derived soils capture and retain
precipitation from rainfall and melting snow without significant lateral distribution in the
landscape. The observed variability in solute concentrations was also small and less than
reported in other field studies. The spatial variability of bromide increased over time
indicating that movement (leaching) differences may be expressed slowly. Paired t-tests
and correlation analyses indicated that the spatial behavior patterns of nitrogen and
bromide were similar in only two of five time periods.
Three types of leaching indices were calculated for both nitrogen and bromide.
Stepwise multiple regression analysis occasionally showed some promise, but the results
were not consistent. None of the three index types produced results more reliable or
consistent than the other two. Exclusion of sites within an ephemeral stream channel also
did not produce substantially or consistently better results. Overall, nitrogen leaching
indices produced slightly stronger relationships with the environmental factors than
bromide indices..
Correlation analysis indicated that wetness indices at two scales (2 m and 10 m)
were moderately correlated to one another and to the thickness of the mollic horizon.
Neither the wetness indices or thickness of mollic horizon were correlated to either
season average measured water content or average measured water content from June of
each year.
These results reiterate the difficulties that are encountered in trying to understand
(predict) the interrelationships between environment, nitrogen leaching, and potential
contamination of ground water.
I
CHAPTER I
SITE-SPECIFIC FARMING, NITRATE LEACHING,
AND GROUND WATER QUALITY
Background
Nitrate Leaching and Ground Water Quality
Ground water is the source of public drinking water for nearly 75 million
people in the United States. Private wells supply water to an additional 30 million
individuals. Evidence indicates that contaminants from agricultural production are
now found in underground water supplies (National Resource Council, 1989). Major
water quality problems of concern to agriculture are nutrients, pesticides, and
sediments. Nutrients and pesticides in water supplies can be a health concern when
levels reach established critical levels.
Non-point sources of pollution have been difficult to identify and regulate
(Bauder et al., 1991), but nitrogen (N) has received the most attention as a threat to
water quality. Nitrite is the most toxic form of N, but children, young animals, and
cattle convert nitrates into nitrites in their stomachs and can develop
methemoglobinemia. In 1962, the U.S. Public Health Service, set an upper limit on
nitrate in drinking water at 10 mg NO3-N per liter (45 ppm nitrate) (McCool and
Renard, 1990). In 1991, Bauder et al. found in a study of private well water quality
2
in Montana that I in every 20 samples had nitrate-N concentrations greater than the
U.S. EPA standard of 10 parts per million. Fifty percent of the high nitrate samples
were attributed to point-source contamination. The other 50% may be the result of
long-term summer fallowing.
Nitrates in ground water originate from several non-point sources, including
geological substrate, septic tanks, improper use of animal manures, cultivation
(especially fallowing), precipitation (acid rain), mineralization of organic N, and
fertilizers (Power and Schepers, 1989). Most nitrate-related environmental impacts
occur at local or regional scales rather than the national scale, and they usually mimic
spatial distribution of the sources (National Research Council, 1978). Conditions
conducive to ground water contamination by fertilizer include shallow ground water,
coarse-textured soils, low organic matter levels, high soil permeability, and excess
water (precipitation, irrigation) (Follett et al., 1991).
Since the late 1960’s, there has been a three-fold increase in applications of N
to Montana soils (Jacobsen and Johnson, 1991). Under native prairie vegetation,
annual N inputs were typically measured in tens of kg ha"1. Today, inputs of several
hundred kg N ha"1 are common with many grain crops (National Research Council,
1978). However, most plants apparently cannot remove all of the NO3-N present in
soil solution. Numerous field studies have shown that less than 50% of the N input
into grain crops is removed in the harvested crop. Therefore, 100 kg N ha"1 or more
is either stored in the soil or lost into the environment (National Research Council,
1978).
3
Nitrate is the end product of the reactions converting other forms of N in soil.
It is stable and remains in soil once it has been introduced, except when removed by
plant uptake, denitrification, or leaching. The quantity of N available to plants is a
function of the amounts applied as manure and fertilizer and mineralized from soil
organic matter. Only a small fraction (usually < I %) of the total N in surface soils is
present as nitrate, but this fraction can undergo rapid changes depending on both
environmental conditions and external sources. Moisture and temperature are major
factors affecting the amount of nitrate in soil, and the nitrate concentrations vary
seasonally, often in a complicated and unpredictable fashion (National Research
Council, 1978). The amount of N released from organic sources, and to some extent,
the amount of fertilizer-derived N existing in the soil after addition of ammonium
(NH4+) or nitrate (NO3") depends on the factors affecting N mineralization,
immobilization, and losses from the soil. Nitrogen fertilizers applied in any form are
converted to nitrate through nitrification. During the growing season, nitrate is
absorbed rapidly by plant roots, probably by a combination of active (metabolic) and
passive (water flow) mechanisms.
Nitrate losses from soil are typically measured utilizing ceramic cups,
lysimeters, collecting water from field drainage, or sampling soil rather than water, or
estimated by computer modeling (Addiscott et al., 1991). The rate of nitrate leaching
from the soil depends on a wide variety of factors. Nitrate leaching from soil is
affected by: plant cover (Hill, 1991), rainfall or irrigation (Thomas, 1970),
temperature (National Research Council, 1978), soil texture and organic matter
4
content (Bergstrom and Johansson, 1991), soil depth, structure and parent material
(Hill, 1991), ground water level and drainage (Hill, 1991), crop strains, tillage
(National Research Council (1978), and forms of N (Hill, 1991).
Nitrate distribution in a soil system is a function of the water balance of the
soil system in that rainfall and irrigation move nitrate downward, while evaporation
moves nitrate back towards the surface. However, the evaporative process is usually
important only in the upper 30 cm of the soil profile (Thomas, 1970). Temperature is
a factor because, water drains more slowly through cold soils. Freezing essentially
stops downward nitrate movement (National Research Council, 1978). Soil texture
and organic matter content can have a major influence on leaching (Bergstrom and
Johansson, 1991). Light and sandy soils tend to retain less water than heavy clay
soils and, therefore, allow more free movement of water and soluble ions following
rainfall. Shallow soils are also usually well drained and are associated with rapid
leaching. Soils rich in organic matter are usually also rich in organic N. This
provides suitable material for the bacterial formation of nitrate and subsequent nitrate
leaching. Tillage stimulates ammonificatipn and nitrification, and most row cropping
practices leave the land bare at least part of the year (National Research Council,
1978). A vegetation-free period and high rainfall can lead to an increased downward
movement of nitrate into aquifers (Hill, 1991).
Spatial Variability of Soil and Water Properties
Most of the soil and environmental parameters which influence the transport of
dissolved nitrates through natural fields vary substantially at different locations, even
5
at short separation distances, and many are not normally distributed (Addiscott and
Wagenet, 1985). Spatial and temporal variability in soil properties with depth in the
soil profile and across landscape positions results in diverse patterns of water and
solute distribution in the landscape. Beckett (1987) points out that the temporal
variability of soil nutrient status may equal or exceed spatial variability..
Hall and Olson (1991) state that two categories of variability, systematic and
random, exist in most landscape studies. Placement in either category may depend on
the level or scale of observation. Soil profile depth, water content, and soil physical
and chemical properties vary with landscape position (Daniels et al., 1985; Brubaker
et al., 1993). This systematic variation of soil properties with landscape position may
affect solute transport under natural rainfall conditions (Afyuni et al., 1994).
Bouma and Finke (1993) alternatively suggest that soil variability may be
attributed to natural soil resource variability or management-induced variability. Soil
resource variability can be expressed in terms of static and dynamic properties. Static
properties are those that, for example, relate to variation in relatively stable soil
properties such as soil texture or organic matter content. Dynamic properties are
those that exhibit differences at short distances; for example, water, solute, air, and
temperature regimes.
Soil textural and structural properties, antecedent soil water content
distributions, and the rate and amount of infiltrating water affect leaching and
movement of water and solutes (Anderson and Bouma, 1977a; 1977b). Even when the
water application rate is relatively uniform over the entire soil surface, such as with
6
rainfall (Jury et al., 1982) or sprinkler irrigation (Van de Pol et al., 1977; Butters,
1987), the resulting vertical water velocities within the soil are not spatially uniform.
Lateral movement can cause substantial variations in solute concentration across a
field of agricultural size (Jury and Nielsen, 1989).
Soil hydraulic properties vary across landscape positions due to changes in soil
profile characteristics (Bathke and Cassel, 1991). Nielsen et al. (1994) state that the
I
soil hydraulic conductivity at water saturation is 10 million times greater than that
exhibited at air-dryness or, stated another way, a seven percent change in soil water
content yields a corresponding 10-fold change in hydraulic conductivity. Therefore,
small differences in soil water content across a seemingly uniform field soil lead to
large differences in the rate of movement of water (and its dissolved constituents)
through the soil.
Site-Specific Farming
Concerns for agricultural sustainability and for protection of surface and
ground water from agricultural chemicals has resulted in increased interest in
precision or site-specific farming. The modem implementation of site-specific crop
management is often thought of in terms of the integration of two complementary
technologies: GPS and GIS. GPS, the satellite-controlled Global Positioning System,
provides the location coordinates for recording data and controlling devices in the
field. GIS (Geographic Information Systems) maintain geographically correct linkages
between GPS-generated data from the field and desktop mapping and decision support
applications (Brown and Saufferer, 1991).
7
In order for a computer simulation model of field-scale leaching to have
practical application in this context, computer programs capable of identifying best
combinations of management practices for a given soil, climate, and water regime
must be developed. Precise guidance of equipment (relative to a previously defined
position) is now possible due to the 24-hour availability of NAVSTAR GPS satellites
and receivers capable of determining position in real time with a resolution of one
centimeter (Larson et al., 1991). In addition to these guidance systems, field
implementation of these management practices benefit from: I) mechanisms (i.e.,
sensors on field equipment) to detect differences within a field with respect to
important soil properties (texture, organic matter, bulk density, water content,
temperature, nutrient status, etc.); and 2) servo-mechanisms controlled by an on-board
computer to alter settings on planters, tillage tools, sprayers, fertilizer spreaders,
cultivators, and other machinery while moving through the field.
Within such a system, a farmer could maintain or enhance crop yield, optimize
fertilizer and other inputs, maximize net return, and also potentially control ground
water contamination or other environmental problems (Power and Schepers, 1989).
Preliminary field experiment results from a soil specific anhydrous ammonia
management system in Minnesota indicate variable benefits, with a maximum of $17
ha"1, when comparing soil specific management to conventional soil management
(Robert, 1991). In 1991, Macy realized a $5.67 ha"1 savings on a 54 ha field in
Wayne County, Indiana utilizing spatially variable control of fertilizer and seed.
8
Carr et al. (1991) used yield and fertility trials in 1987-88 at five locations to
demonstrate that fertilizing by soil is profitable under dryland systems in Montana.
On-Farm. Field-Scale Information Systems
Many technologies necessary to implement site-specific crop management
practices are currently being developed as separate and narrowly focused, dedicated
applications. Brown and Saufferer (1991) state that unless these "closed subsystems"
can be networked into a whole-farm, integrated management information system, their
potential value will be greatly limited.
There is a growing demand for simple techniques applicable for day-to-day
land management decisions, but which also incorporate environmental protection
considerations. To that end, integrated technological systems including on-farm,
field-scale information systems and site-specific farming provide opportunities for
"farming soils, not fields" (Carr et al., 1991). Implementation of integrated
agricultural management practices can have two major consequences: I) effects on
the environment including ground water quality; and 2) effects on economic return
from the agricultural enterprise (Power and Schepers, 1989).
Field-Scale Solute Transport Models
Approaches used to model the transport of a mobile solute such as nitrate
through soil can be divided into two groups: deterministic models and stochastic
models (Tables I and 2) (Addiscott and Wagenet, 1985). Predictive models are
designed for estimating the effects of changing inputs or transfer rates on pool sizes,
9
leaching losses, and related issues (Winteringham, 1984). Knowledge of two aspects
is needed for a complete description of leaching: I) the position of the chemical’s
peak concentration in the soil profile (i.e., solute center of mass); and 2) the shape of
the leaching band (Smith et al., 1984).
Deterministic models develop a description of transport based on mass
conservation and flux laws, and rely on differential equations to predict values of the
water and solute variables as functions of position and time. In contrast, stochastic
models describe the variables as random functions, which depend on the distribution
of soil property values, to predict concentration averages and variances. These
statistics are used to calculate the probability of having a given value appear at a
given depth or time (Jury and Nielsen, 1989).
Deterministic models require detailed measurements of transport and rate
coefficients for predicting nitrate movement through soil. The deterministic piston
flow model provides only a rough estimate of nitrate movement, but can be applied
using only minimal information about soil properties (Jury and Nielsen, 1989).
However, the spatial variability of water and solute transport parameters in natural
field soils has made it difficult to apply detailed deterministic models to field-scale
solute transport. In 1985, Jury reviewed field-scale solute transport experiments and
reported coefficients of variation of 60-200% for apparent solute velocity. Jury
(1982) and Jury et al. (1986) proposed a stochastic, non-mechanistic solute transport
model using a transfer function to represent solute transport and reactions implicitly.
10
Table I. Review of leaching models (after Addiscott and Wagenet, 1985).
DETERMINISTIC
• presume system or process operates such that occurrence of
given set of events leads to uniquely-defined outcome
MECHANISTIC
FUNCTIONAL
• define rates of change
• usually based on rate parameters
• driven by time
• incorporate fundamental process
mechanisms
• based on convection-dispersion
equation
• serve primarily as research tools
• define changes
• usually based on capacity parameters
(less spatially variable than rates)
• driven by rainfall, evaporation,
irrigation
• incorporate simplified treatments of
solute and water flow
• no claim to fundamentally
• require less input data and computer
expertise
• input data obtainable independently of
test data
• more widely used as management
guides
Analytical Solution
• assume steady-state solute movement,
steady-state one-dimensional water flow,
homogeneous soil
• general form, flexible incorporation of
processes
• widely usable if experiments simulated
meet required boundary conditions
• mainly restricted to controlled,
homogenous, de-structured lab columns
• main value is validating conceptual
framework
• Problems:
1. constraints o f boundary conditions
required rarely represent field conditions
2. some constants needed must be
obtained by fitting model to test data
3. subject to problem of variability
Partially Analytical
• assume piston flow, compute solute
peak position, impose effects of
dispersion and diffusion around peak
• displacement calculated using water
content at field capacity, not rates of
movement
• potential use as both research and
management tools, but with limitations in
each case
• Problems:
1. cannot handle heterogeneous profiles
or those which contain solute at depth at
beginning of leaching period
2. do not address immobile water
associated with intra-aggregate pores
11
Table I continued. Review of leaching models.
Numerical Solution
Layer & Other Simple Approaches
• most widely used
• useful for management purposes
• solid mathematical basis
• mathematically simple
• solve convective-dispersion equation
• modest input requirements
for non-steady-state, transient water and
• no prior fitting procedures required
solute conditions
Chromatographic
• use finite difference and finite element
• relevant to exchanged/adsorbed solutes
methods
• flexible in range o f initial and
Layer
boundary conditions for water and solute
• offshoots of chromatographic approach
• avoid site specificity by appropriate
• divide soil into horizontal layers
use o f soil physical and chemical data
• accommodate mobile and immobile
• Problems:
1. susceptible to problems o f variability water phases
• handle non-uniform initial solute
2. modeled solute and water fluxes
tested against measured water content in distributions and multiple solute pulses
•Problem arises if inputs fail to simulate
previous validation exercises
3. predictions misleading unless inputs important variability in real leaching
well characterized in terms of variability
Mass balance or conservation
• several equations required
• computer program needed
STOCHASTIC
• presuppose the outcome to be uncertain
• allow solute and water movement to vary spatially with soil properties
• mainly used in research due to shortage o f suitable field studies for validation
MECHANISTIC
NON-MECHANISTIC
• use randomly-distributed input values
to produce a distribution of output values
• useful for assessing impact of
variability
• use transfer or probability density
functions
• consider soil to be composed of
twisted capillaries o f different lengths
within which water moves by piston flow
• simulate spatially-variable field
processes with minimum input data
• unknown if satisfactory in vertically
inhomogeneous soils or if give accurate
estimates of flux as well as concentration
12
Table 2. Models for solute leaching in soils (after Addiscott and Wagenet (1985) and
Pennell et al. (1990)).
DETERMINISTIC
MECHANISTIC
FUNCTIONAL
Analytical Flow Solution
Partially Analytical
Nielsen & Biggar (1962)
Nielsen et al. (1973)
Biggar & Nielsen (1976)
Van Genuchten & Wierenga (1976)
DeSmedt & Wierenga (1978)-Piston flow
Scotter (1978)-Soil voids
Rose et al. (1982 a,b)-Mass conservation
Nofziger & Homsby (1987)-CMLS
Steenhuis et al. (1987)-MOUSE
Numerical Flow Solution
Laver & Other Simple Aooroaches
Bresler & Laufer (1974)
Tillotson & Wagenet (1982)
Carsel et al. (1984)-PRZM
Parker & VanGenuchten (1984)-CXTFIT
Wagenet & Hutson (1987)-LEACHM
Knisel et al. (1989)-GLEAMS
Sharma & Taniguchi (1991)
Chromatoeraohic Aooroach
Levin (1964)
Frissel & Poelstra (1967)
Thomas et al. (1978)
Preferential Flow
Beven (1981); Beven & Germann (1981)
Bouma et al. (1981)
Trudgill et al. (1983)
Laver
Terkeltoub & Babcock (1971)
Bums (1972, 1974, 1975)
Addiscott (1977)-mobile/immobile water
Nichols et al. (1982)
Addiscott (1982)
Mass Conservation
Bresler (1967)
Tanji et al. (1972)
STOCHASTIC
MECHANISTIC
Biggar & Niesen (1976)
Dagan & Bressler (1979)
Bresler & Dagan (1979, 1981)
Amoozegar-Fard et al. (1982)
Scaline Theory
Bresler et el. (1979)
Wagenet & Rao (1983)
NON-MECHANISTIC
Probabilitv Densitv/Transfer Functions
Raats (1978)
Jury (1982), Jury et al. (1982, 1986)
White et al. (1986)
Kachanoski et al. (1992)
Destouni (1993)
13
The solute transport properties of a volume of soil between the soil surface and
outflow surface (i.e., some specified depth where solute is monitored) is represented
by a solute travel time probability density function (Jury and Nielsen, 1989). Jury
and Nielsen (1989) indicate that the advantages of this approach for solute transport in
the field are: I) the formulation of transport is general enough to encompass all
linear solute transport paths; and 2) the transfer function may be calibrated without
measuring the soil water transport properties directly.
Amoozegar-Fard et al. (1982) used an alternative stochastic, mechanistic
transport model for spatially-variable field soils. The Monte Carlo model assumes
that a specific process model (such as the convection-dispersion model) is valid at a
specific location, but that the model parameters have a distribution of values over the
field. Simulations are run by drawing values for the parameters randomly from their
probability distributions. The field-averaged solute concentration and its variance is
calculated from the set of individual outcomes using the random values of the
parameters (Jury and Nielsen, 1989).
Afyuni et al. (1994) state that predictive models for solute transport in soil
must consider the landscape if they are to accurately reflect the process as it occurs in
nature. The empirical, predictive approach utilized in this project incorporates
consideration of presumed topographical controls on soil water through the use of
terrain attributes relating to water flow in and across the landscape.
14
Scope and Purpose
The work presented in this thesis is part of a larger effort to examine the
possibility of integrating site-specific information about leaching into an on-farm,
field-scale information system.
Integrating soil survey, terrain, and microclimate
attributes and image analysis along with information regarding leaching potential into
a GIS format may provide new opportunities for applying leaching potential models to
variable soil landscapes. Subsequent identification of areas susceptible to leaching, or
management zones which consider leaching susceptibility as an important factor,
would provide a basis for more selective N fertilizer application. Site-specific
applications of N fertilizer may improve profitability, and at the same time, reduce
the possibility of ground water contamination. This approach is currently being
attempted on a regional level. Wylie et al. (1994) are developing and validating
methods of linking the Nitrate Leaching and Economic Analysis Package (NLEAP)
(Schaffer et al., 1991) to a GIS framework to identify and map regional nitratenitrogen leaching distributions under irrigated agriculture in the western U.S. The
NLEAP model uses farm management, soil, and climate information to estimate
nitrate leaching indices. The overall purpose of the work presented here was to
investigate the relationships between field-scale nitrate leaching and soil, soil water,
and topographic-characteristics in a cultivated Montana farm field.
15
Objectives
Two projects relating to field-scale leaching are described in this thesis. The
first project, discussed in Chapter 2, encompassed two objectives: I) to characterize
and describe the spatial and temporal variability of nitrate and bromide leaching in a
20-ha Montana farm field of moderate relief (43 m); and 2) to test the leaching
prediction capabilities of several soil, soil water, and terrain attributes that singly, or
in combination, represent the spatial distribution of soils and microclimate attributes
in this farm field under uniform application of N and bromide. Chapter 3 describes
the second project to determine if a topographically-derived wetness index calculated
at two grid scales (2 m and 10 m) is correlated to field-measured soil water content
(collected over three field seasons including two cropped and one unusually wet,
fallow year).
16
CHAPTER 2
DESCRIPTION AND PREDICTION OF FIELD-SCALE LEACHING
j
Introduction
The degradation of water quality from agricultural sources is a critical concern
for both surface and ground waters in the U.S. (Crowder and Young, 1988).
Pollution prevention practices are more cost effective than after-the-fact clean-up
practices. Reclaiming contaminated ground water is impossible or uneconomical in
f
many instances (McCool and Renard, 1990). There is evidence that nitrates already
accumulated deep in the soil profile will raise ground water contamination levels even
if corrective agricultural practices are adopted immediately (Winteringham, 1984).
More than 50% of Montanans (95% in rural areas) depend on ground water as a
source of drinking water (Jacobsen and Johnson, 1991; Algard et al., 1994) and
legislation has been enacted, to protect this resource. The Montana Agricultural
Chemical Ground Water Protection Act, which took effect on January I, 1990,
encourages proper management of agricultural chemicals to prevent, minimize, or
mitigate their presence in ground water (Hart and Jacobsen, 1994).
The links between agricultural land use and ground water quality are not as
•V
well understood as those for surface water. Most agricultural pollution is non-point in
nature. Estimating the potential for ground water contamination by agricultural
\
17
chemicals is complicated by complex geological, hydrological, and biological
processes (Crowder and Young, 1988).
In the 1980’s, soil N was identified as the single most important factor
contributing to problems of "increased productivity with simultaneous resource
protection" (Winteringham, 1984). Nitrogen fertilizers are the most frequently cited
cause for nitrate accumulation in ground water, although other sources can
collectively contribute hundreds of kilograms of NO3-N ha"1 each year (Power and
Schepers, 1989). Nitrate is a mobile form of soil N and can reach ground and
surface water by several mechanisms including surface erosion, runoff, local reabsorption of volatilized N compounds, or leaching transport mechanisms.
In agricultural landscapes, ground water contamination occurs in areas where
surplus soil water leaches excess nitrates or other soluble compounds below the crop
foot zone and potentially to ground water. Although nitrate is taken up rapidly by
growing plants, this removal mechanism only occurs within the crop root zone. The
nitrates may be derived from fertilizers, legumes, manure, or natural soil sources.
Failure to match crop nutrient needs with soil, water nutrient supply may lead to the
presence of excess nitrates in the soil system when: I) residual (surplus) nitrates
from over-fertilized areas are present; or 2) lack of plant-available N decreases water
use and promotes percolation beyond the root zone.
Nutrient management on the farm can reduce agricultural non-point pollution.
Use of known best management practices (BMPs) has been demonstrated to be highly
effective in controlling leaching of nitrates (Power and Schepers, 1989). However,
18
the control of nitrate leaching is complicated. Reduction of surface runoff and its
constituents through implementation of soil conservation practices may increase
percolation of water through the soil root zone and subsequently recharge of ground
water aquifers, thereby potentially increasing the leaching of nitrates, pesticides, fecal
coliform bacteria, and phosphorus to ground water. Because fertilizers are the
predominant N source in agricultural systems, the most effective adjustments in
management practices needed to control ground water degradation will involve
fertilizer practices (Crowder and Young, 1988).
Changes can be made in fertilizer practices to enhance fertilizer use efficiency
and thereby reduce potential for nitrate leaching. The general relationship between N
fertilizer application rates and crop yields is qualitatively well known. The growth of
a crop in a given location and efficiency of the crop’s use of available N, however,
depend on soil properties, weather and climate, and cultivation and management
practices. The complex interaction of these factors makes it extremely difficult to
predict exact fertilizer needs for a specific crop at a specific location, or alternatively,
to determine the amount of residual N lost to the environment (National Resource
Council, Commission on Natural Resources, 1978). Synchronizing the amount and
timing of N fertilizer and other nutrient applications with the crop’s requirement is
probably the most cost-effective and efficient method to control pollution in surface
and ground water (Crowder and Young, 1988; Addiscott et al., 1991). Consequently,
uniform fertilizer management is a potential source of ground water contamination.
Uniform management creates inefficiencies by over-treating and under-treating
19
portions of fields. This can increase field management cost, waste energy, decrease
net return, and potentially contribute to both surface and ground water pollution
(Burrough, 1993). The spatial variation of soil properties, terrain, and microclimate
causes uneven patterns in soil fertility and crop growth. This variation decreases
fertilizer use efficiency when applied uniformly at the field scale (Miller et al., 1988;
Bhatti et al., 1991; Larson and Robert, 1991; Mulla, 1993). Soil spatial variability
within fields has been identified by soil testing and crop yield differences. Detailed
soil surveys or crop-free aerial photographs show that individual fields often include
several soil types (Robert and Rust, 1982).
Recent technologies in geographic soil information systems, remote sensing
techniques, simulation modeling techniques, and computerized field applicators offer
new opportunities for "farming soils, not fields" by applying fertilizer selectively to
take account of the variation in fertility and leaching capacity of the soil (Carr et al.,
1991; Larson and Robert, 1991; Burrough, 1993). If the spatial variation of soil
fertility and leaching potential over a field could be mapped, fertilizer applications
could be.adjusted based on management zones, thereby reducing excess fertilizer
application. A Global Positioning System (GPS) receiver and a small computer with
a digital map of management zones in the cab of a tractor is required. This allows
the amount of fertilizer released from application equipment to be adjusted according
to each management zone as the tractor progresses through a field (McEachem et al.,
1990; Petersen, 1991). The cost savings in fertilizer inputs and reduction in ground
water pollution may be sufficient to justify the capital and time expenditure.
20
Practical and operational (field-usable) solutions to ground water contamination
problems will require consideration of contrasting environmental differences within
fields. Many mathematical expressions and models have been developed to predict
the downward leaching rate and depth-concentration profile of solutes. One difficulty
in designing leaching process models is that when the model successfully simulates a
particular soil profile, it may not be representative of the overall field situation due to
spatial and/or temporal variability. Additionally, water infiltrating down the soil
profile can bypass uniform displacement mechanisms by entering structural or
biological voids in the soil (Winteringham, 1984; White, 1985).
New techniques and models as well as additional data may be needed to
predict field-scale leaching (Mulla, 1990; 1991; 1993). Model and data requirements
to achieve this goal must be reasonable if the model is to be used for farm planning.
Many studies attempting to relate soil properties or crop parameters to landscape
position have used general landscape mapping units (hilltops, midslopes, footslopes)
to characterize soil position rather than specific topographic attributes (Walker et al.,
1968; Stone et al., 1985; Brubaker et al., 1993, 1994; Er-Raji and Cihacek, 1994).
Two types of topographic attributes are significant with respect to characterizing the
spatial variability of the landscape: I) primary attributes (including slope and aspect)
which are calculated directly from digital elevation models (DEM); and 2) secondary
attributes or compound topographic indices which may help describe or characterize
spatial variability of specific processes occurring in the landscape, such as soil water
distribution (Moore et al., 1993a; 1993b).
)
21
The integration of soil survey (soil attributes and boundaries)^ terrain (threedimensional topography), microclimate (precipitation and transpiration), and image
analysis (precise pest and crop performance mapping) into a GIS format would add
detail to soil management areas within an on-farm GIS (He et al., 1992; Mausbach et
al., 1993; Robert, 1993; Wilson et al., 1994). Management areas could then be
identified based, in part, on their leaching susceptibility. Fertilizers could then be
precisely applied within a field with GPS receivers guiding application equipment.
More selective N fertilizer application practices based on soil, terrain, microclimate,
\
and imagery information may reduce dr avoid ground water contamination, and
improve profitability.
Objectives
This study encompassed two objectives: I) to characterize and describe the
spatial and temporal variability of nitrate and bromide leaching; and 2) to test the
leaching prediction capabilities of several soil, water, and terrain attributes that
singly, or in combination, represent the spatial distribution of soil and climatic indices
in a farm field under uniform application of N and bromide. The results could
potentially provide a tool to be used by farmers in making land management decisions
for ground water protection.
/
22
Description of Study Area
The study area consisted of a 20 ha (50 acre) portion of a farm field located at
the base of the Bridger Mountains (2700-2950 m) in the Gallatin Valley near the
community of Springhill, Montana (TIN, R6E, Section 18; 45°50’40" 111°2’12")
(Figure I). It is bounded on the south by a gravel county road and on the east by a
gravel driveway/road. In general, it has a southerly aspect, moderately strong relief
(43 m), and an average elevation of 1509 m (Figures 2 and 3). A small intermittent
stream runs through the field in a south-south-westerly direction (see bottom half of
Figures 2 and 3). The field has been dryland farmed by the Wright family with a
grain-fallow rotation for approximately 50 years.
The underlying geologic materials are soft tertiary valley fill sediments,
alluvium, and loess derived from Devonian to Cambrian rocks (Veseth and Montagne,
1980). The soils in this field are dominated by: I) Fine-silty, mixed Typic
Argiborolls (Farland silt loam) formed in loess and stratified alluvium on terraces and
valley footslopes; and 2) Coarse-silty, mixed Typic Ustochrepts (Brodyk silt loam)
associated with the Farland silt loam. Some Fine, mixed Argic Cryoborolls (Bridger
variant clay loam) formed in very thick loam or clay loam unconsolidated deposits
from a variety of igneous, metamorphic, and sedimentary rock deposited as alluvium
or glacial till in intermountain valleys were also present (National Cooperative Soil
Survey, 1994).
23
24
Figure 2. Elevation contour map (Spangrud, 1995).
10
o
10
20
30
40
50
□
E
W —rW
METERS
I
25
Figure 3. Interpolated 10 m elevation lattice (Spangrud, 1995).
I mesh cell = 5 by 5 meters
W
26
Farland silt loams, were found on nearly, level to gently sloping terraces and
footslopes of stream valleys (4-8% and 8-15%), are deep, well-drained soils that
formed in stratified alluvium of mixed mineralogy. Brodyk silt loams consist of very
deep, well-drained soils formed in loess on dissected stream terraces with slopes
ranging from 8 to 15 % and were found in association with Farland silt loams.
Bridger variant clay loam, which was found on a steep slope (15-35%), consists of
deep, well-drained soils formed in colluvium and residuum from limestone and shale.
All three soil series exhibit medium to slow/moderately slow runoff and moderate
permeability (National Cooperative Soil Survey, 1975; 1978; 1992).
The climate at Gallatin Field (45°47’N 111°09’W, 1353 m) located
approximately 14 km southwest of the study area in the Gallatin Valley is cool, semiarid, with long cold winters (Ruffner, 1985). However, the study site experiences
cooler and wetter conditions due to its location next to the base of the Bridger
Mountains. The study site experiences a mean annual air temperature of about
6.7 °C (44 °F) (Caprio et al., 1994). Most of the precipitation falls as snow and
spring rain. The study area receives mean annual precipitation of about 41 to
76 cm’ (16 - 18 inches) and mean annual snowfall of about 127 to 254 cm (50 - 100
inches) (Caprio et al., 1994).
Material and Methods
The following sections describe field management during the course of this
study, field data collection, sample analysis, generation of terrain attributes, data
preparation, and the statistical methods used for data analysis.
27
Field Management
The timeline followed in this study is summarized in Table 3 and additional
details are discussed below by year.
1992. Initial soil samples were collected and described during installation of
seventy neutron access tubes. Additional soil samples were collected for neutron
probe calibration. Potassium bromide (KBr) solution was then surface applied as a
tracer to test plots adjacent to each neutron access tube. The KBr solution was
prepared from pellets and applied using a CO2 pressured backpack unit equipped with
a fan spray nozzle. A constant volume of 0.5 liters KBr solution (50.37 g KBr L'1)
was placed in a plastic 2-liter soda bottle and delivered across each plot as evenly as
possible. The KBr solution was sprayed inside a 1.5 m by 1.5 m quadrat for an
application rate of 150 kg ha"1 (133.9 lbs ac"1). Grain yield was sampled both by hand
and by plot combine around each sample site.
1993. Barley stubble remained in the field over the winter of 1992-1993 and
the field remained fallow until early October when it was seeded with winter wheat
CTriticum aestivum L.). In July, Roundup solution was sprayed on each KBr test plot
and around each neutron access tube to prevent uptake of nitrogen and bromide by
weeds. Roundup was applied using a CO2 pressured backpack unit equipped with a
spray nozzle. In late September, KBr solution was re-applied to test plots adjacent to
each neutron access tube and soil samples were again collected. The KBr solution
28
Table 3. Springhill project timeline.
FALL 1991
Sep
GPS data collected and sample grid of 70 sites established
SPRING 1992
May 14-20 Initial soil samples collected, field descriptions of soil profiles
recorded, and neutron access tubes installed
May 15
Field sprayed with Roundup herbicide
Jun I
Field seeded at ~ 67-101 kg ha'1 barley IHordeum vulgare L.)
and fertilized at ~ 56-22-28-22 kg ha"1 (50-20-25-20 lbs ac"1) N-P-K-S
Jun 5
Surface application of bromide (KBr) solution on test plots at
150 kg h a 1 (~ 133.9 lbs ac'1)
FALL 1992
Sep 29/Oct I Grain yield and plant biomass samples collected
Oct. 19-20 Field harvested
Oct. 21-23 Soil samples collected
SPRING 1993
May I
Neutron access tube locations resurveyed
May 25/Jun 4 Soil samples collected
I % Roundup herbicide solution sprayed around each access tube and on
Jul I
each bromide test plot
Field tilled and sprayed with Roundup
Jul 1-2
Field tilled
Jul 14
Field tilled
Aug 3-4
FALL 1993
Surface application of bromide (KBr) solution on test plots at
Sep 23
337.5 kg ha'1 (-3 0 1 .3 lbs ac'1)
Soil samples collected
Sep 28-29
Riser sections of neutron access tubes removed from field and lower
Sep 30/
section sealed below ground level
Oct I & 5
Field
fertilized
at — 90-34-34 kg ha'1 (80-30-30 lbs ac'1) N-P-K
Oct 10-11
Field seeded at —101 kg ha"1 (90 lbs) winter wheatITriticum aestivum L.)
Oct 18
and FarGo applied at rate of —14.5 kg ha"1 (12.94 lbs ac"1)
Neutron access tubes relocated with hand-held metal detector
Oct 22-23
SPRING 1994
Apr 16-17
Riser sections of neutron access tubes replaced
Apr 15-20
Field tilled
Apr 18-19
Soil samples collected
May 2
Field reseeded at — 84 kg ha"1 (75 lbs ac"1) barley IHordeum vulgare L.)
FALL 1994
Aug 20-24 Field harvested
Sept I
Field tilled
Sept 1-3,6
Soil samples collected
Oct 12
Neutron access tubes removed from field
29
was sprayed inside a I m by I m quadrat for an application rate of 337.5 'kg ha"1
(301.3 lbs ac"1). After planting, two sample points were not able to be reestablished
(#20 and #48).
1994.
Neutron access riser tubes were replaced in mid-April while the field
was being tilled to prepare for planting barley CHordeum vulgare L.) due to the
failure of the winter wheat. The field was re-seeded after re-installation of the
neutron access tube risers and the equipment paths that were followed to avoid
crushing the neutron access tubes led to non-uniform seeding and the occurrence of
weeds around the tube sites and across the bromide test plots. Therefore, plant
samples were not collected for either grain yield determination or plant biomass and
uptake.
Preparation of GPS Survey arid Sample Grid
Elevation data were collected at 6,284 different locations in Fall 1991 with an
Ashtech Sensor GPS receiver mounted on a pickup truck and an Ashtech P-12 GPS
receiver operating in kinematic mode at a fixed reference location. (Figure 4). This
survey was performed by Dr. David Tyler from the Dept, of Surveying Engineering
at the University of Maine-Orono. The data collected included longitude and latitude
in state plane meters (NAD 1983) and elevation values in meters above sea level. A
permanent sampling grid consisting of 70 sample points was established in the field
(Figure 5). A 5 .1-cm (2-inch) PVC neutron access tube was installed at each sample
point to 200 cm below the soil surface using a truck-mounted hydraulic soil sampler
and 5 .1-cm (2-inch) diameter bit. Each neutron probe access tube consisted of two
30
Figure 4. GPS collection pattern (Spangrud, 1995).
IO
o
10
20
30
40
J W
E
METERS
a
31
32
sections. The lower section was installed with the top edge at least 20.3 cm
(8 inches) below ground level. Two 1.3 cm (0.5-inch) holes were drilled in the lower
section 3.8 cm (1.5 inches) below the top of the tube and covered with electrical tape.
The purpose of these holes was to facilitate extraction of the tubes at a future date. A
22.9 cm (9-inch) piece of rebar was buried on the north side of each lower section
tube to aid in relocation. A riser section of the same diameter was fitted over the
lower section through the use of a collar or sleeve to extend the tube assembly above
ground level for neutron probe access. The riser section was sealed with a 5.1 cm
(2-inch) solid rubber stopper (Size 11-1/2) between neutron probe data collections.
The locations and elevations of the neutron access tubes were resurveyed in
May 1993 by Mr. Brian Autio and his students from the Dept, of Civil and
Agricultural Engineering at Montana State University utilizing an HP-48 data collector
with a Topcon GTS-303 total station.
Soil Sampling
Initial soil samples were collected and described in May 1992. Field
descriptions and measurements of the mollic epipedon, Bt, Bk, and 2C horizons,
color, texture, and percent clay were obtained from the cores removed for neutron
access tube installation. The cores were then divided into 15 cm and 30 cm
increments to depths of 30 cm and 210 cm, respectively, for laboratory analysis. Soil
samples were collected in lined, paper sample bags and placed in coolers in the field
for transportation back to Montana State University where they were air-dried,
ground, and screened through a 2-mm sieve prior to analysis. Soil testing performed
33
on these initial samples included pH (1:2 solid isolation ratio), EC (1:2 dS m"1), OM
(%) determined by the Walkley-Black method, extractable Olsen P (mg kg'1), and
extractable K (mg kg'1) for 0-15 cm, and extractable NO3-N (mg kg'1) for each depth
increment to 210 cm. All soil and plant analyses were performed by the Montana
State University Soil Analytical Lab.
Soil samples were subsequently collected each spring and fall for three years
from the KBr test plot to a maximum depth of 180 cm (in the increments listed above)
with a hydraulic soil sampler utilizing a 3.4-cm (1.34 inch) diameter bit. The
samples were handled and processed as described above before analysis for both
nitrate-N and bromide (mg kg'1).
The soil samples were analyzed for NO3-N via a colorimetric cadmium
reduction method after a 1:5 soil to solution calcium hydroxide extraction. The
procedure used for analysis of soil bromide was developed at MSU by Dennis Hengel
(Appendix A).
Generation of Terrain Attributes
The terrain attribute data Were generated by Damian Spangrud. The following
description is paraphrased from his unpublished thesis, and a more detailed
description of the procedure may be obtained there (Spangrud, 1995). All of the GPS
elevation data were converted to a regular, rectangular 10 m DEM grid using
ANUDEM (Hutchinson, 1989). ANUDEM uses a series of partial, thin-plate splines
with drainage enforcement to create grid-based DEMs from irregular point or contour
data.
34
The interpolated 10 m elevation lattice data (Figure 3) were then entered into
the grid version of the Terrain Analysis Programs for the Environmental Sciences
program (TAPES) (Moore, 1992) to calculate terrain attributes for each 100 m2 cell in
the study area. Five of the twelve primary terrain attributes generated were used in
the current study: slope, aspect, profile (down-slope) curvature, plan (across-slope)
curvature, specific catchment area (area draining to a cell divided by the width of that
cell). Additionally, one secondary or compound topographic attribute (steady-state,
relative wetness index) was calculated with the WET program (Moore, 1992).
TAPES-G allows different attribute prediction algorithms to be used based on
user preference for topographically important variables such as slope and specific
catchment area. Slope was calculated using the FD (finite difference) approach rather
than the D8 (approximate) approach. Specific catchment areas can be calculated in
TAPES-G using either the D8, RhoS (a stochastic version of the more common D8
/
algorithm), or FRhoS algorithms. The FRhoS algorithm was utilized in this study
because it allows flow to be : I) routed to any of the eight neighboring cells on a
slope-weighted basis; and 2) distributed to multiple cells in upland areas above the
channel head. Modeling of upland non-stream areas was important to this project
because there is no defined stream channel system over most of the study area.
The steady-state wetness index was calculated with the WET program and is
derived from two primary TAPES-G attributes, specific catchment area and slope as
follows:
w =
ln(As/tan /3)
(I)
35
where w indicates the wetness index, As is the specific catchment area (m2m'1), and /3
is the slope angle (degrees). Two important assumptions are incorporated into this
wetness index: I) that the direction of subsurface flow (dictated by the gradient of
the piezometric head) is parallel to the gradient of the surface topography; and 2) that
steady-state conditions apply. Equation (I) provides a relative index where the value
increases with increasing specific catchment area and decreasing slope gradient.
Larger indices indicate cells with low slope and large drainage areas relative to other
cells in the study area. Therefore, hilltops (flat areas with low specific catchment
area) exhibit moderate values, valleys (high specific catchment area and low slope)
where water concentrates exhibit high values, and steep Mllslopes (high slope) where
water drains more freely exhibit low values (Moore et al., 1993a; 1993b).
The terrain attributes calculated by TAPES-G were then imported into
ARC/INFO (Environmental Systems Research Institute, Inc., Redlands, CA 92373)
and a point-in-polygon overlay of the 67 sample point coordinates and the grid was
performed to determine cells in which the sample points were located. The terrain
attributes for each cell containing a sample point were then extracted from
ARC/INFO as an ASCII file for further analysis.
Three sample points (#52, #64, #65) were removed from analysis when the
DEM coverage was clipped to a rectangular area. Subsequently, two of these sample
points (#64 and #65) were independently lost due to the construction of a new house
on the west side of the field during the study.
36
Soil Water Content
Neutron probe (Gardner, 1986) data was collected in 25-cm intervals between
25 and 200 cm depth at each sample point using a CPN Hydroprobe moisture meter.
Measurements were taken throughout the growing season for three years as follows:
I) approximately 10-day intervals from 6-8-92 to 11-12-92; 2) approximately 15-day
intervals from 5-18-93 to 9-24-93; and 3) approximately 10-12-day intervals from
4-21-94 to 8-17-94, Water which periodically accumulated in the bottom of some
x.
neutron access tubes was removed as necessary via a suction apparatus. This problem
was more prevalent in the spring and probably resulted from rain entering the tubes
due to cap removal or loss between data collections.
Wet and dry calibration soil samples and corresponding neutron probe readings
were collected in May 1992 and September 1992, respectively at 11 locations across a
range of topographic positions. The resulting probe calibration equation is:
Y = 0.20X - 0.013
(2)
where Y is the predicted volumetric water content (Oy) and X is the neutron probe
count ratio (measured count -f- standard count). This regression yielded an R2 value
of 0.65 which was somewhat lower than expected.
Equipment problems resulted in standard count readings varying from about
8000 to 10,000 on different days of data collection. A standard count of
approximately 10,000 is expected, thus, all count ratios collected on monitoring dates
with a standard reading of less than 10,000 were adjusted to a 10,000 standard count
J
37
before further calculations were made. The count ratios recorded at each depth for all
sites for each monitoring date were then transformed to depth equivalent of water
(cm) by:
D6 = 0VD
(3)
where 6Vis the measured volume water content and D is the corresponding
measurement depth interval in centimeters.
From this transformed data, the average total water content (cm of water) in
the soil profile was calculated over each growing season for each site for three
different total profile depths (25 cm, 100 cm, and 150 cm).
Grain Yield and Plant Uptake Measurements
Barley grain yield and plant biomass samples were collected in September
1992. Grain yield was determined by hand clipping and plot combine methods.
Hand collected grain samples were obtained from the northeast side of each neutron
access tube sample point by clipping barley heads from an area I m in length across 4
rows (25.4 cm (10 inches) apart) giving a sampling Cell of 1.02 m2. These grain
samples were threshed and grain yield (kg m"2) was determined by weighing each
grain sample on a toploading balance.
Separate barley grain yield samples were collected from the four quadrants
around each sample point using a plot combine which employed a 1.2 m header
giving a sampling cell of 0.006 ha. Barley grain yield for the plot combine Samples
was determined by weighing each of the samples on a large capacity beam balance in
38
the field. An average yield (kg nr2) for each sample point was then calculated by
averaging the four sample cells surrounding a sample point.
A small sub-sample of barley grain was taken at the time of Fall 1992 harvest
for test weight and protein determination. Percent protein was determined by the
Montana State University Cereal Quality Laboratory using infrared reflectometry
techniques.
A 0.305 m * 0.254 m (12" * 10") area (sampling cell of 0.077 m2) of
barley was also clipped at ground level and collected from the KBr test plot for
determination of plant uptake. The samples were weighed on a toploading balance to
determine plant biomass (g m'2), ground, and then analyzed for N and bromide.
The Kjeldahl procedure was used for analyzing the plant tissue for total N (%)
(Bremner and Mulvaney, 1982). The procedure for bromide (%) analysis of plant
tissue was adapted by Dennis Hengel at MSU from the bromide selective-ion
electrode method reported by Abdalla and Lear (1975) (Appendix A).
Barley grain yield and plant biomass samples were not collected in Fall 1994
due to variable seeding rates around neutron access tubes.
Computations for Leaching Characterization
A minimum of two leaching indices were generated for both N and bromide
for each sampling period. Where appropriate, four additional indices were generated.
The two primary indices were generated using a BASIC program written with the
assistance of Dr. Jon M. Wraith in the Dept, of Plant, Soil, and Environmental
Sciences at Montana State University.
39
The LEACHIND.BAS program was created to generate the two primary.
Teaching indices (Figure 6; Figure 14 in Appendix B). This program calculates the
total solute mass and a "critical depth" of solute within the soil profile from smoothed
data. The program begins by fitting an n-1 order polynomial function to the
measured solute concentration by depth data at each site and then interpolating data at
5-cm depth increments from the fitted equation. The next step is numerical
integration of the fitted, interpolated data at each site using Newton’s forward (2nd
order), Bessel’s (3rd order), and Newton’s backward integration (2nd order)
procedures.
Each of these procedures is based on finite difference approximations to
the integral. Lastly, the program calculates two leaching indices for each site: I) the
total solute load (ie. solute mass) in the soil profile to a depth of 180 cm; and 2) a
depth marking the boundary above which 80% of the total solute load is found (ie.
critical depth). The program was rerun twice to also calculate a critical depth for
25% and 50% (i.e., center of mass) of the total solute load. Field sites that were not
•
I
sampled at a particular time period were treated as missing values. Additionally, two
secondary leaching indices were calculated from the solute mass and critical depth
values, i.e., positive or negative change in these values from one sampling date to the
next were calculated for each applicable time period.
Where applicable, two other leaching indices were calculated based on
conservation of mass estimates. For each applicable time period, a mass balance
leaching index was calculated at two depths, 60 and 90 cm. Sixty centimeters was
selected because it is the common rooting depth of barley (Sims, 1994). The ninety
40
Figure 6. LEACHIND.BAS flowchart.
DECLARE
subroutine and functions
------------------- 7
INTERPOLATE
data at 5-c m increments
from fitted equation
S i/
DEFINE
i n p u t / o u t p u t d a ta files
SET
all interpolated data points bwt last
two measured points to zero, for cases
where last two measured points were
equal to zero
OPEN
i n p u t / o u t p u t files
\/
READ
FlagJD1 Depth, Concentration
CONVERT
d e p th values fro m in c h e s to c m
V/
CALCULATE
TSolMass (m g /k g )
a n d 80% C ritD epth (cm )
PRINT
re su lts to o u tp u t files
41
J
• .
centimeter depth was chosen for comparison as it is the greatest depth to which most
sites were sampled in any one time period.
Three sources of N exist during the growing season. One comes from residual
N in the soil, the second comes from mineralization and nitrification of organic
matter, and the third is from fertilizer applied. The additions and removals from the
system included in the mass balance calculations for both nitrogen and bromide during
each time period are listed in Table 4. Denitrification as a nitrate removal reaction
through biological transformation from soil was considered negligible in this study
area and was not included in the mass balance calculations. Denitrification primarily
occurs under low oxygen conditions and is most likely to take place in moist soils
with high water holding capacity or inside aggregates which are water saturated
(Broadbent and Clark, 1965).
The only input factor not accounted for was the possible reintroduction of N or
bromide onto the soil surface with decaying plant biomass. Quantitative information
on N or bromide release from plant residue was not available, and therefore, was not
included as an addition in the calculations.
The mass balance calculations were performed on a pounds per acre basis.
Bulk density values were not available, so all data were transformed to lbs ac"1 units
via the weight of an acre furrow slice. It is customary to consider that an average
acre furrow slice (6.67 inches) has a bulk density of 1.3 g cm"3 and weighs about
2,000,000 pounds or 1,000 tons (Brady, 1990). Due to using this uniform
transformation across the range of topographic sites and with depth, some information
about variability was not incorporated. The soils were sampled in 15 cm (6-inch)
42
increments, therefore (6 4- 6.67 inches)*(2,000,000 lbs of soil) yields 1,799,100 lbs
per 6-inch acre furrow slice.
This corresponds to an assumed uniform bulk density
of 1.32 g cm"3.
Table 4. Mass balance worksheet factors.
1992
Nitrogen
Soil
1994
1993
Spring
Fall
Spring
Fall
Spring
Fall
+3
+
+
+
+
+
Fertilizer
+
Mineralization1
+
Plant Uptake23
+
+
+
+
-
+
-
Bromide
+
Soil
Plant Uptake2
+
+
+
-
+
-
1 Nitrogen mineralization assumed to be 30 lbs/ac/yr/% organic matter.
Values pro-rated for time interval between sampling dates.
2 Rooting depth for barley at 60 cm (Sims, 1994).
3 " + "represent addition;
represents removal.
A benchmark soil in Gallatin Valley, the Amsterdam series, a Fine-Silty,
Mixed Typic Cryoboroll, is similar to the soils present at the study site. Bulk density
values (1/3 bar) for this soil range from 1.40 g cm"3 in the surface horizon (0 - 16
cm) to 1.25 g cm'3 at the greatest depth measured (135 - 180 cm) (benchmark soil
samples at MSU). The assumed bulk density used in the data transformations is,
therefore, equivalent to the average bulk density found in the Amsterdam series.
43
For each time period, an equation was developed to set N or bromide leaching
(i.e., % of solute unrecoverable) equal to the sum of other appropriate terms listed in
Table 4. For example, in Fall 1992 the percent unrecoverable N was calculated as:
[(Unrecoverable N) 4- (N Present in System)] * 100
(4)
or:
[(N Present in System - N Recovered in Plant and Soil Samples)
4- N Present in System] * 100
(5)
or specifically:
[(SP92 N level + Fertilizer + Mineralization - F92 Plant Uptake - F92 N level)
4- (SP92 N level + Fertilizer + Mineralization)] * 100
(6)
The results from the mass balance calculations indicate the percent of solute
unrecoverable at the current ,sampling date across a common sampling depth of 60 or
90 cm in the soil profile relative to what was present at the previous sampling date.
Mass balance indices were not calculated for the last time period, Spring 1994 to Fall
1994, as barley samples were not collected for determination of plant uptake due to
non-uniform seeding and occurrence of weeds on the test plots. Negative percentages
(results) were treated as missing values.
Statistical Analysis
All edaphic, terrain, soil water, and yield variables along with the leaching and
mass balance indices were stored for each of the 67 sample points in database files.
Tables 43 and 44 in Appendix C present complete lists of the initial baseline values
for the measured soil and both measured and predicted terrain variables, respectively.
44
These data were exported to SAS (SAS Institute, Inc., 1988) as ASCII data files to
create permanent SAS data files to facilitate correlation and regression analyses.
Correlation analysis was performed to estimate correlation coefficients for each
pair of independent variables, dependent variables (N and bromide) and the dependent
and independent variable combinations. Paired t-tests were performed between the
solute mass and mass balance indices for N and bromide for each time period using
MSUSTAT (Lund, 1993).
Stepwise multiple regression analysis was used to separate those environmental
factors (edaphic, terrain, soil water, and yield) which correlated best with N and
bromide leaching. The dependent variables (leaching indices) for each time period
were considered one at a time in each regression procedure. The exact number of
independent variables utilized in any one regression procedure differed based on the
presence or absence of a crop and other relevant factors for each sampling period.
The F level for entry or deletion of an independent variable was set to 0.05.
Statistical significance of the overall equation was determined by an F-test. The t-test
was used to test the significance of each independent variable. The models were also
evaluated for over- or under-specification using Mallow’s Cp value. Plots of
regression residuals were examined to determine the presence of outliers.
Results and Discussion
Leaching Behavior
Nitrogen. Table 5 lists the values of average N solute mass in the soil profiles
integrated to a depth of 180 cm from the LEACHIND.BAS program which are
graphically depicted in Figure 7. In S92 at the beginning of field monitoring, an
average of 459.1 mg kg"1 was present in each soil profile. The distribution of the N
was such that 80% of the N mass was present within the top 103 cm, 50% within 60
cm, and 25% within 34 cm of the soil surface (Table 6; Figure 8). This amount and
distribution was then considered as "time zero" or baseline for subsequent N
calculations.
Table 5. Summary statistics for nitrogen solute mass1 (mg kg'1).
Variable
S92
F92
n
Min
67
41.0
65
17.0
Max
1873.1
Mean
S.D.
C.V.
SE*2
459.1
332.0
0.72
40.57
523.9
115.7
SE x/X 3
0.09
100.9
0.87
12.52
0.11
S93
S94
F93
66
67
3 2 .7
68.3
776.5
1345.5
227.1
147.3
0.65
18.13
0.08
421.6
244.8
0.58
29.90
0.07
1 To a depth of 180 cm.
2 Standard Error of the mean.
3 Standard Error of the mean as a proportion of the mean.
65
218.2
4179.3
1297.4
753.6
0.58
93.47
0.07
F94
64
2 .3
1127.4
258.0
255.3
0.99
31.92
0.12
46
Figure 7. Average nitrogen solute mass (mg kg*1).
1400.00
1200.00
1000.00
800.00
600.00
400.00
200.00
- -
Fertilizer
A ddition
Season
From S92 to F92 a barley crop was present in the field. Analysis of soil
samples collected after barley harvest showed that the average N solute mass
decreased from 459.1 to 115.7 mg kg*1 (Table 5; Figure 7). However, since N
fertilizer was added to the field with barley seeding in the spring after the soil
samples had been collected, the maximum N concentration was not captured.
Therefore, the decrease shown is less than that experienced in the field. The N
unrecoverable in the plant or soil samples (Table 7; Figure 9) was 56.4% to 60 cm
and 46.6% to 90 cm depth (roughly similar to the percent unrecoverable for bromide
during the same time period). However, the critical depths for the remaining N
decreased for all three levels, although by differing amounts. The critical depths
containing both 80% and 25% of total solute decreased by similar amounts from
J
Table 6. Average nitrogen critical depth (cm).
PTSL*
F92
n = 65
S92
n = 67
Sample
25
50
80
25
50
25
80
50
80
25
%
%
%
F93
n = 67
S93
n = 66
50
S94
n = 65
80
25
%
50
F94
n = 64
80
25
%
50
80
%
Min
10
25
40
5
5
10
5
10
20
5
15
40
5
10
15
5
5
5
Max
80
105
150
105
115
150
75
145
155
85
115
150
100
no
135
11
13
15
Mean
34
60
103
19
57
85
29
56
91
20
48
101
20
41
83
35
66
98
S.D.
23
49
34
29
27
42
C.V.
22
58
37
29
35
43
SEx
2.83
6.07
.4.22
.3.51
3.31
5.29
SExZX
0.03
0.07
0.05
0.03
0.04
0.05
* PTSL = percent of total solute load
48
103 to 85 cm and 34 to 19 cm respectively, but a much smaller change in the critical
depth at 50% was observed, decreasing only slightly from 60 to 57 cm (Table 6;
Figure 8). Although N fertilizer was applied to the field at the time of barley
planting in S92, plant uptake and N mineralization in surface layers of the soil
probably led to the observed decreases in the critical depth values at all levels.
Figure 8. Average nitrogen critical depth (cm).
120.00
100.00
.g 6o.oo
40.00
Season
25% — 50%
80% I
Over the fallow winter months from F92 to S93, average total N mass
increased to 227.1 mg kg '. In conjunction, the nearly parallel line segments for F92S93 in Figure 9 indicate that the percent of N unrecoverable at both 60 and 90 cm
both increased by virtually the same increment, 22.9 and 22.6%, to 79.3 and 69.2%
respectively.
Although the 50% critical depth did not change (57 to 56 cm), the
49
critical depth for both the 25% and 80% levels both increased from 19 to 29 cm and
85 to 91 cm, respectively (Table 6; Figure 8).
Table 7. Summary statistics for nitrogen balance (% unrecoverable).
Variable
n
Min
Max
Mean
S.D.
C.V.
SEx
SE5ZX
F92
60 cm 90 cm
62
47
25.8
9.8
78.5
99.9
56.4
46.6
12.3
0.22
1.56
0.05
S93
60 cm 90 cm
60
40
18.8
9.8
92.4
89.5
69.2
79.3
15.6
12.8
18.2
0.33
2,28
0.10
0.16
1.65
0.04
0.26
2.88
0.10
F93
60 cm 90 cm
50
42
0.9
0,9
87.8
83.1
42.6
43.2
23.1
0.54
3.26
0.17
S94
60 cm 90 cm
54
58
0.2
0,2
82.4
78.4
37.6
20.8
44.3
17.1
0.48
3.21
0.20
0.39
2.25
0.10
0.46
2.36
0.13
17.4
During the summer fallow period from S93 to F93, the average N mass
increased from 227.1 to 421.6 mg kg"1 (Table 5; Figure 7). This was expected due to
increased mineralization of N from organic matter and increased breakdown or decay
of crop residue during the warm months. This increase was larger than that observed
for F92 to S93. Simultaneously, the N balance for both 60 and 90 cm decreased
substantially, converging on nearly the same percentage, 42.6% and 43.2%,
respectively (Figure 9). The N distribution data presented in Figure 8 concurs with
both the N solute mass and mass balance data. The data show that 25% and 50% of
the N in the profiles was closer to the surface by the F92 sampling. The critical
depth decreased at both levels by virtually the same amount, 9 and 8 cm respectively.
The fact that the percent unrecoverable and critical depth both decreased makes sense
50
in light of N mineralization in surface mollic horizons. However, the critical depth
for the 80% level increased from 91 to 101 cm. The N present in the lower profile
moved downward with summer precipitation. The summer of 1993 was the coolest
and wettest on record (Montana Climate Center, 1995).
Figure 9. Average nitrogen balance (% unrecoverable).
100.00
80.00
60.00
40.00
20.00
Season
60 cm
90 cm
After soil sampling in F93, N fertilizer was applied to the field prior to
planting of winter wheat. Subsequently, the S94 sampling revealed a large increase in
average N mass in the profiles to 1297.4 mg kg'1 (Table 5; Figure 7). Figure 9
shows that some of the N, residual and applied, moved partially downward through
the profile over the winter months. The percent of unrecoverable N slightly increased
in the top 60 cm and slightly decreased in the top 90 cm. This information along
51
with the S94 critical depths in Figure 8 indicate that applied and residual N in the
profiles moved downward relative to the distribution in F93. The critical depth for
the 80% level decreased by almost 20 to 83 cm while the critical depth decreased by
7 to 41 cm for the 50% level and remained the same for the 25% level at 20 cm.
After S94 soil sampling, the field was reseeded with barley due to the failure
of the winter wheat. The percent N unrecoverable was not calculated for F94 since
plant biomass samples were not collected to monitor plant uptake of N or bromide.
Soil analysis revealed that the average N mass in the profiles decreased sharply from
1297.4 mg kg'1 in S92 to 258.0 mg kg'1 after harvest in F94 (Table 5; Figure I).
Although N uptake by barley plants undoubtedly occurred, Table 6 and Figure 8
indicate that the average critical depth of the N increased by 15 cm at the 25% and
80% levels, while the 50% critical depth level increased more dramatically by 25 cm.
This is the opposite trend seen in the S92 - F92 cropped period where critical depths
decreased. The S92 sampling followed a fallow period and much more N was present
in the soil profiles (4179.3 mg kg'1). It appears that excess N not taken up by the
barley crop leached downward.
Bromide. In field experiments, bromide ion has been widely used as a proxy
or tracer to indicate how nitrate moves in soils. Bromide ion is a suitable tracer for
solute transport in many soil systems for three reasons: I) its natural background
concentration in most waters and soils is very low, thus movement of small amounts
can be detected (Kung, 1990; Onken et al.,, 1977; Agus and Cassel, 1992); 2) it can
be easily analyzed (Onken, et al., 1975); and 3) it is nonreactive (Kung, 1990).
52
Several studies have shown that nitrate and bromide behave similarly in the subsoil
(Smith and Davis, 1974; Onken et al.,.1977). However, bromide cannot be used to
determine quantitative movement of nitrate (Onken et al., 1977) , because it is not
subject to processes involved in the soil N cycle. In other words, bromide is a
biologically conserved tracer and cannot be used to determine "specific questions
regarding the fate of fertilizer N in a soil-plant system" (Silvertooth et al., 1992).
Silvertooth et al. (1992) also indicate, however, that the concentration profile of
bromide can "infer solute movement patterns and present a worst-case scenario
regarding potentials for nitrate leaching loss."
In laboratory column studies without root systems, bromide ,has been shown to
be. a good tracer for the movement of nonreactive solutes and its breakthrough
patterns can be well described by a convection-dispersion model (Clothier and Elrick,
1985; Jardine et al,, 1988). However, the uptake of bromide ions from the root zone
and reintroduction back to the soil surface can significantly change the character of a
bromide pulse under certain conditions and with specific plants (Kung, 1990; Wraith
and Inskeep, 1994). Therefore, the dependability of using the bromide front to
indicate the velocity of solute movement in a field soil with active roots, and whether
the bromide breakthrough curve (BTC) can represent the fate of some contaminants
that can be degraded in plants needs to be addressed (Kung, 1990).
Table 8 lists the summary statistics for bromide solute mass in the soil profile
integrated to a depth of 180 cm for each sampling. Figure 10 graphically depicts the
change in average bromide solute mass over time listed in Table 8.
)
1
53
Table 8. Summary statistics for bromide solute mass (mg kg'1).
Variable
S92
F92
S93
F93
S94
F94
n
Min
67
0
65
31.1
66
0.0
67
0.0
65
51.3
64
8.4
Max
0
0
644.7
197.6
137.1
0.69
17.00
0.09
1189.1
173.2
238.8
1.38
29.39
0.17
3145.0
325,5
702.2
2.16
85.78
0.26
2631.3
980.2
542.9
0.55
67.34
0.07
1413.0
Mean
S.D.
C.V.
SEx
SEx/X
383.3
276.2
0.72
34.53
0.09
As expected, average bromide total solute mass increased between S92 and
F92 from zero to 197.6 mg kg'1 due to the addition of KBr solution (150 kg ha"1 or
133.9 lbs ac"1) after the S92 soil sampling. However, by F92, 53.1% of the surfaceapplied bromide was unrecoverable to 60-cm, depth and 39.3% was unrecoverable to
90-cm depth after accounting for uptake of bromide by the barley crop present within
the bromide test plot during the 1992 growing season (Table 9; Figure 11). Although
these data appear to provide strong evidence for bromide leaching during the first
growing season, it is also possible that some plants adjacent to the bromide test plot
took up some of the applied bromide. It is also possible that lateral movement of
bromide occurred due to concentration gradients. Table 10 and Figure 12 indicate
that 50% of the bromide in the soil was within the top 19 cm and 80% was present in
the top 46 cm. In other words, over half of the bromide applied is unrecoverable in
the top 60 cm, however, 80% of the remaining bromide is present in the top 46 cm of
the soil profile.
54
Figure 10. Average bromide solute mass (mg kg'1).
S 600
g 400
200 T-
KBr
A pplication
KBr
Application
Season
1st Application
2nd Application
Although Figure 10 (Table 8) indicates that the average total solute mass of
bromide in the soil profiles decreased only slightly from 197.6 to 173.2 mg kg'1
during the F92 to S93 period, Figure 11 (Table 9) reveals that amount of bromide
unrecoverable increased at nearly the same rate from 53.1 to 86.4% to 60 cm and
from 39.3 to 68.9% to 90 cm. Table 10 and Figure 12 show that the depth to which
the soil must be sampled to recover 25 and 50% of the remaining bromide increased
at nearly the same rate from 8 to 58 cm and 19 to 72 cm, respectively. The critical
depth for 80% increased at a slightly lower rate from 46 to 87 cm. These results
indicate that the bromide remaining in the profiles after the F92 barley harvest moved
significantly downward, but remained within 180 cm of the soil surface.
55
Additionally, 11 sites were identified from the S93 sampling where the bromide had
moved sufficiently down the profile such that the solute was deeper than sampling
depth of 180 cm. At the other sites, only a small amount of bromide was leached
from the sampled soil volume, but significant downward movement of the remaining
bromide was observed during this winter fallow period.
Table 9. Summary statistics for bromide balance (% unrecoverable).
F92
Variable
n
Min
Max
Mean
S.D.
C.V.
SEx
SE x/X
S93
60 cm
90 cm
60 cm
90 cm
62
13.0
82.4
47
0.9
75.0
56
43.4
53.1
16.7
0.31
2.12
0.04
39.3
18.5
0.47
2.70
0.07
32
2.3
100.0
68.9
28.2
0.41
4.99
0.07
100.0
86.4
14.9
0.17
1.99
0.02
F93
60 cm 90 cm
60
100.0
100.0
100.0
0
0
0
0
52
0.0
100.0
98.1
13.9
0.14
1.92
0.02
S94
60 cm 90 cm
64
23.8
98.7
68.1
17
0.25
2.12
62
19.7
98.4
65.1
17.9
0.27
2.27
0.03
0.03
Between S93 and F93 the field was fallow and significant downward
movement Of bromide was again observed, but was accompanied by an increase in
average solute mass. Figure 10 (Table 8) indicates that the average solute mass
almost doubled from 173.2 to 325.5 mg kg'1. Although 42 sites contained no bromide
within the sampling depth of 180 cm, ten values from the F93 sampling were greater
than 1000 mg kg"1and two of those ten values were greater than 3000 mg kg'1. These
high values skewed the mean value for bromide solute mass.
56
Figure 11. Average bromide balance (% unrecoverable).
100.00
2 60.00
D 40.00
20.00
Season
1st App-60cm
1st App-90cm
2nd App-60cm
2nd App-90cm
The second application of KBr was surface applied on the test plot immediately
prior to the F93 soil sampling. By F93, most of the bromide left in the soil profiles
resided at the deeper depths sampled. Therefore, prior to calculating the F93
leaching indices, the bromide concentration values reported by the Soil Analytical Lab
were adjusted by subtracting out the known amount of bromide just applied on the
surface in an attempt to reveal an actual decrease in bromide solute mass from the
first KBr application. After this adjustment, some sites still displayed surface
concentration values of bromide greater than zero. This is very likely due to bromide
redistribution to the soil surface as a result of a mechanism by which bromide is taken
up by plants and later redistributed to the soil surface as observed in barley and
canola (Wraith and Inskeep, 1994). Kung (1990) speculated that this might occur
Table 10. Average bromide critical depth (cm).
F92
n = 65
PTSL3
25
50
80
25
SEx
SEx/X
5
60
8
.
5
130
19
80
25
10
145
46
35
76
4.38
0.09
5
120
58
50
80
25
%
%
%
Min
Max
Mean
S.D.
C.V.
50
S94
n = 65
F932
n = 25
S931
n = 55
5
135
72
10
150
87
34
39
5.59
0.08
1 In S93, bromide deeper than 180 cm at 11 sites.
2 In F93, bromide deeper than 180 cm at 42 sites.
3Percent total solute load.
40
155
122
50
160
132
50
F94
n = 64
80
25
%
65
165
144
23
16
8.72
0.16
5
45
19
10
80
31
50
80
%
15
5
120
10
58 - 18
25
43
3.06
0.05
5
12
44
5
15
80
37
46
4.63
0.06
58
when dead leaves decay. Owens et al. (1985) showed that, in a field study, as much
as 30% of applied bromide can be taken up by grass and reintroduced into the soil
surface after the grass has decayed.
Figure 12. Average bromide critical depth (cm).
Season
1st App.-25%
1st App.-50%
1st App.-80%
7nrl App.-25% -s- 2nd App.-50% - 3 - 2nd App.-80%
However, since the total increase in bromide mass is larger than can be
accounted for by this mechanism, it is probable that some measurement or calculation
error may be involved. The expected overall decrease of bromide solute mass is
supported by the data in Figure 11 (Table 9) showing that the percent of
unrecoverable bromide to both 60 and 90 cm increased, converging toward a value of
100.0% (98.1% to 90 cm depth). In addition, 42 sites (compared to 11 in S93) were
identified where the bromide was no longer present within the sampling depth of 180
59
cm. At the other 25 sites, the critical depth of the bromide remaining in the profiles
increased almost linearly from 58 to 122 cm, 72 to 132 cm, and 73 to 144 cm at the
25, 50, and 80% levels respectively (Table 10; Figure 12). These data indicate only
very small amounts of bromide present in the 0 - 90 cm portion and a significant
amount of bromide now located between 90 and 180 cm.
As mentioned above, a second application of 337.5 kg ha"1 (or 301.3 lbs ac"1)
KBr was surface applied in F93. This application is reflected in the large increase in
average total mass of bromide solute to 980.2 mg kg'1 in the S94 sampling (Table 8;
Figure 10). Over the F93 to S94 winter period, winter wheat (which ultimately failed
in S94) was present in the field. By the S94 sampling, nearly 70% of the bromide
was unrecoverable (68.1% to 60 cm and 65.1% to 90 cm) (Table 9; Figure 11). This
is a much larger first period increase in percent unrecoverable than observed
following the first KBr application in S92. These data indicate that either more
leaching or lateral water flow occurred because the field was effectively fallow
and/or there was some plant uptake by the winter wheat crop during this period. The
critical depths for recovery of 25, 50, and 80% of the bromide were 19, 31, and 58
cm respectively Table 10; Figure 12). These changes in critical depths during a
winter period are greater than those observed for the summer cropped period from
S92 to F92 directly following the first KBr application. It is interesting to note,
however, that the distances between the critical depth at each level are virtually
identical; 11-12 cm between the 25 and 50% levels, and 27 cm between the 50 and
80% levels. Therefore, although different amounts of solute were present in the soil
profiles, the relative distribution is the same.
60
From S94 to F94, average bromide mass in the soil profiles sharply decreased
by nearly two-thirds from 980.2 to 383.3 mg kg'1 (Table 8; Figure 10). Two types of
losses were possible during this period, leaching and plant uptake and the data reflect
a large decrease in bromide solute mass. The percent of bromide unrecoverable was
. not calculated for F94 because no plant samples were collected as discussed earlier
(p. 38). The rate of increase in critical depth (Table 10; Figure 12) at all three levels
is less (-1 vs. 8 cm at 25% level; 13 vs. 19 cm at 50% level; and 22 vs. 46 cm at
80% level) than that measured during the 1992 growing season following the first
application of KBr. However, the 1994 growing season was the driest of the three
included in this study which may explain the slower downward movement.
Nitrogen Behavior vs. Bromide Behavior. Paired t-tests were performed
between N and bromide mass by season to determine whether the spatial patterns
were similar. Table 11 shows that the t-test result was highly significant on three of
the five sampling dates (F92, S94, and F94) indicating that spatial patterns of N and
bromide variability were different. In S93 following a winter fallow period, the
t-test was not significant at the 5% level, but was significant at the 10% level
(P = 0.0724) indicating some differences in spatial behavior patterns. After two
fallow periods in F93, the t-test was highly insignificant (P = 0.3064) indicating that
N and bromide behavior was more similar during the summer of 1993. Pearson’s and
Spearman Rank correlation coefficient values (Table 11) varied from |r | = 0.3023 to
0.4040 at the 5 % level of significance. Although the relationship between N and
61
bromide solute mass is relatively weak based on r-values, it is definitely notable that a
significant relationship is present for three of the five time periods.
T ab le'll. Comparison of nitrogen and bromide solute mass by season.
Paired
t-test
Pearson’s
Correlation
Spearman Rank
Correlation
n
P-value
r-value
P-value
r-value
P-value
Fall 1992
65
0.0000
-0.1455
0.2496
-0.1835
0.1435
Spring 1993
66,
0.0724
0.3023
0.0136
0.2592
0.0356
Fall 1993
67
0.3064
-0.0873
0.4821
0.1025
0.4094
Spring 1994
65
0.0000
0.3347
0.0064
0.2469
0.0474
Fall 1994
64
0.0010
0.4040
0.0000
0.4037
0.0000
Sampling
Date
Statistical comparisons of the N and bromide mass balance estimates were also
performed by season (Table 12). The t-test results to 60 cm depth were highly
significant on three of the four sampling dates (S93, F93, and S94). In F92 following
a cropped period, the results were not significant to 60 cm depth (P = 0.2432). The
t-test to 90 cm depth was highly significant on half of the sampling dates (F93 and
S94). Although, the t-test to 90 cm depth for F92 was not significant at the 5% level,
but was significant at the 10% level (P = 0.0681). However, at the S93 sampling
date (following a winter fallow period) the results to 90 cm depth were not significant
(P = 0.7223) indicating a higher degree of similarity in N and bromide behavior over
f
62
the winter 92-93 fallow period. Review of the Pearson’s and Spearman Rank
correlation data (Table 12) again reveals mostly weak relationships between N and
bromide mass balance figures. The only significant (5 % level) r-values obtained
occur for the Spearman Rank correlation in F93 to both 60 and 90 cm depths.
Table 12. Comparison of nitrogen and bromide mass balance by season.
Paired
t-test
Pearson’s
Correlation
Spearman Rank
Correlation
n
P-value
r-value
P-value
r-value
P-value
Fall 1992
61
0.2432
-0.0097
0.9408
0.0370
0.7769
Spring 1993
56
0.0021
0.2551
0.0578
0.1774
, 0.1908
Fall 1993
48
0.0000
0.0000
1.0000
0.5445
0.0000
Spring 1994
58
0.0000
0.0322
0.8105
0.0553
0.6802
Fall 1992
46
0.0681
0.0626
0.6792
0.0735
0.6276
Spring 1993
32
0.7223
0.0492
0.7892
-0.1194
0.5152
Fall 1993
40
0.0000
0.0606
0.7103
0.5051
0.0000
Spring 1994
54
0.0000
-0.0202
0.8847
-0.0156
0.9111
Sampling
Date
60 cm
90 cm
The data presented in Tables 11 and 12 are not surprising in that one would
not expect N and bromide to behave similarly in the field. Bromide is a non-reactive,
63
biologically conserved substance. Once bromide is applied on the surface, it is either
taken up by plants or moves through the profile with soil water. The original amount
applied may be tracked as no new sources of bromide exist in the soil. In contrast, N
is not only taken up by plants, but is also involved in N cycling interactions.
Additionally, a new source of N exists via mineralization near the soil surface.
These complexities make tracking fertilizer N applied difficult.
Conclusion. Evaluation of N and bromide behavior through examination of
three indices (solute mass, critical depth, and mass balance) show that leaching of
both N and bromide did occur in this field. In general, the data seem to make sense
in light of N and bromide applications and the environmental factors involved in each
time period. It is interesting to note that greater variation in critical depth values for
N was not observed across the five sampling periods. Nitrogen has a source for
addition of new material via mineralization near the surface, while bromide does not.
Paired t-tests between N and bromide solute mass and % unrecoverable yielded mixed
results regarding the similarity of spatial behavior patterns. The solute mass
comparisons seem to indicate that N and bromide behaved most similarly when the
field was fallow from S93 to F93. However, the mass balance comparisons (%
unrecoverable) indicate greatest similarity during the S92 to F92 cropped period and
over the following winter fallow period from F92 to S93.
Environmental Controls
Data representing environmental factors that potentially explain nitrate leaching
were gathered at each of the 67 sites. These data were grouped into measured
64
edaphic, water, and biotic (plant), and both measured (x coordinate, y coordinate, and
elevation) and computed topographic categories. Tables 13 through 16 present the
sample numbers, minimum and maximum values, means, and standard deviations of
the edaphic, topographic, water, and biotic factors, respectively.
Table 13. Summary statistics for edaphic controls.
n
Min
Max
Mean
S.D.
C.V.
pH
(1:2)
EC
(1:2)
(dS nr1)
67
5.6
8.3
6.3
0.6
0.09
67
0.04
0.23
0.07
0.03
0.43
OM
(%)
K
(mg kg"1)
Olsen
P
(mg kg"1)
Thickness
of Mollic
Horizon
(cm)
67
166
844
318
121
0.38
67
0.0
93.1
43.6
19.0
0.44
67
10
126
45
27
0.60
67
1.08
5.55
3.47
0.91
0.26
Depth
to
CaCO3
(cm)
67
10
>180
Table 14. Summary statistics for topographic controls.
Elev
(m)
Slope
(%)
Aspect
(deg)
Spec;
Catch.
Area
(m2 nr1)
Wetness'
Index
Plan
Curv.
(m nr1)
Profile
Curv.
(m n r1)
n
Min
67
1489.85
67
1.503
67
64.295
67
10.8
67
4.533
67
-18.245
67
-1.844
Max
Mean
S.D.
C.V.
1532.08
23.259
10.054
5.187
0.52
332.319
191.228
57.170
0.003
8717.5
592.2
1849.3
3.12
13.271
6.940
1.771
0.25
28.838
0.463
8.474
18.30
1.187
1508.89
12.10
0.01
0.081
0.471
5.81
65
Table 15. Summary statistics for soil water controls: average total water (cm).
1992
25
cm
n
Min
Max
Mean
S.D.
C.V.
SEx
SE%/X
67
5.64
7.88
6.77
0.59
0.09
0.07
0.01
1993
1994
100
cm
150
cm
25
cm
100
cm
150
cm
25
cm
100
cm
150
cm
65
14.41
25.50
18.24
1.71
0.09
0.43
0.02
64
25.90
55.98
32.78
3.88
0.12
65
6.79
9.39
8.36
0.61
0.07
0.95
0.19
0.02
61
22.56
29.19
26.02
1.23
0.05
0.93
0.04
57
23.35
40.00
36.89
2.81
0.08
0.18
65
5.52
8.14
6:55
0.54
0.08
0.18
61
19.14
25.15
21.49
1.17
0.05
0.77
43
29.07
34.58
32.42
1.31
0.04
1.92
0.05
0.03
0.04
0.09
0.03
Table 16. Summary statistics for biotic controls: Fall 1992 barley yield and plant
uptake.
Grain
Hand
Collection
(kg nr2)
n
Min
Max
Mean
67
0.078228
0.580837
0.336014
S.D.
0.080379
0.24
C.V.
Yield
Plot
Combine
(kg nr2)
66
0.035879
0.415944
0.278093
0.067889
0.24
Plant
Biomass
(g)
Plant
Nitrogen
(%)
67
59.15
175.52
100.65
19.92
67
0.44
1.81
0.90
0.27
0.20
0.30
Plant
Bromide
(%)
67
0.28
2.02
0.74
0.29
0.39
Observed Spatial and Temporal Variability
Agus and Cassel (1992) state that a common field experience is that substantial
variability exists in both the depth of solute movement and its concentration even
when extreme care is taken to apply the chemical uniformly over the soil surface.
66
However, results from the current study indicate only modest spatial variability in
both soil water content and solute measurements.
Soil Water. Table 15 lists descriptive statistics for average total water content
(cm) measured to three depths during each growing season. The consistent small
coefficient of variation values (CV) (0.04 - 0.12) and standard error ratios (SE^/X)
(0.01 - 0.09) for each depth within each growing season indicate the lack of
substantial spatial variability in water content across the 70 sites or with depth.
The lack of spatial variability in water content is surprising in that the study
area exhibits topographic relief which should theoretically translate into redistribution
of water across the landscape. However, in three field seasons spent collecting
neutron probe data (including the wet, fallow 1993 season), no evidence of surface
runoff was observed other than in the ephemeral stream channel. A possible
explanation of this lack of spatial variability in water content is the possibility that the
downward component of water movement overwhelms any tendency towards lateral
subsurface flow in these deep, permeable agricultural soils (Wraith, 1994). The data
in Table 15 appear to support this explanation. The season averages for 1993 were
not substantially higher than those for 1992 or 1994 to any depth indicating that water
is lost quickly or moves rapidly in this environment or that most values were near
field capacity, especially below the zone of substantial root uptake.
Undoubtedly some temporal variability of water content was not incorporated
by calculating season averages for use in multiple regression. Difference and
correlation comparisons of the season averages to the water content measured in June
67
(wettest date near the beginning of each season) of each corresponding year were
performed to evaluate differences and to determine whether the season averages
retained similar patterns of variability.
Average differences between the season averages and June data to each of the
three depths are shown in Table 17. In 1992, the differences between the season
average water content values and those measured in June were all negative.
Specifically, the season average total water content values to 25, 100, and 150 cm for
each site were on average 3.19, 13.24, and 11.83 cm, respectively, drier than the
measured total water content at the same site in June. The CV values were 0.11,
0.11, and 0.28 for the three depths. In 1993, small positive and negative differences
occurred between the season averages and values for June. The season average total
water content values to the three depths were on average 0.18 and 0.15 cm wetter to
the 25 and 50 cm depths, respectively, and 0.09 cm drier to the 150 cm depth than
the measured water content values in June. The CV values of the differences were
1.40, 1.78, and 10.28, respectively. In 1994, the differences were again all negative
and the CV values were 0.87, 0.29, and 0.42 respectively for the three depths. The
season average total water content values were on average 1.09, 2.70, and 2.91 cm
drier than those measured at the corresponding site in June.
Table 18 shows that both the Pearson’s and Spearman Rank correlation
coefficients between the season average and June data were fairly high. The
correlations were executed on a data set which included all sites (Stream Channel In
= SCI) and on a smaller subset which excluded the sites lying within the ephemeral
68
stream channel (Stream Channel Out = SCO). The sites considered to be in the
stream channel were 30, 31, 34, 42, 43, 47, 55, and 69. These sites receive
additional water from outside the study area via a culvert located on the east side of
the field (Figures 2 and 5). The r-values obtained for the SCO data set were slightly
higher than for the SCI data set.
Table 17. Differences between soil water season averages and June data.
1992
1994
1993
25 '
cm
100
cm
150
cm
25
cm
100
cm
150
cm
25
cm
100
cm
150
cm
65
-13.24
1.48
56
-11.83
3.26
60
0.15
0.27
50
-0.09
0.97
C.V.
0.11
0.11
0.28
64
0.18
0.25
1.40
1.78
10.26
56
-1.09
0.95
0.87
50
-2.70
S.D.
64
-3.19
0.34
41
-2.91
1.22
0.42
n
Mean
0.78
0.29
The high r-values from Pearson’s correlations indicate that the general pattern
of relationships of water content between sample sites is relatively strongly reflected
in the season averages. The high Spearman Rank correlation coefficients indicate a
high degree of similarity between the ranking of the season averages and June data
from wettest to driest. However, paired t-tests between the season average water data
and the June data were highly significant (P = 0.0000) for each combination for both
SCI and SCO; with one exception. The P-value for the season average-June pair to
150 cm in 1993 was 0.4984 (SCI) and 0.9176 (SCO). These results indicate that the
spatial patterns of variability of the season average water data were significantly
different from the June data except to a depth of 150 cm in the wettest year, 1993.
69
The data presented in Table 17 show that the soil water season averages
differed from the June data by the largest margins in 1992. However, the variation of
these differences were relatively small (as indicated by the CV data), with the
exception of the values for 150 cm in 1993 (CV > 10). The data presented in Table
18 reveal that the soil water season averages are highly correlated to the
measurements taken in June of each corresponding year. The lower r-values obtained
for the 100 and 150 cm depths in 1992 concur with the larger differences shown in
Table 17. Finally, the paired Hest results indicate that with the exception of the 150
cm depth increment in 1993, the season average and June values do not share similar
patterns of variability. Therefore, although the season average water values share a
general relationship with the June .data, their distribution across the landscape differs.
Table 18. Correlations for total soil water between season averages and June data.
n
Stream Channel In
Spearman
Pearson’s
r
r
64
64
0.8520
0.9132
0.8436
0.8851
65
0.6722
1992
65
1993
1994
150 cm
1992
1993
1994
25 cm
1992
1993
1994
n
Stream Channel Out
Spearman
Pearson’s
r
57
r
0.9050 .
0.9243
0.7248
56
57
0.9029
0.9399
0.7499
0.6337
0.8154
58
0.7387
0.8519
60
58
0.9755
0.8006
0.9648
0.7590
54
0.9750
0.8000
0.9666
0.7637
56
50
41
0.4903
0.8759
0.7455
0.7437
0.9492
53
47
0.7912
0.9564
0.7546
39
0.4760
0.9408
0.7568
0.7548 "
100 cm
53
0.7623
.
70
Solutes. The spatial variability of resident solute mass concentrations of
nitrate and bromide in the soil profiles at each sampling date are listed in Tables 5
and 8, respectively. These tables show small CV and SB^/X values for N at any
sampling date and small CV and SEx/X values for bromide that increase with time
after application.
The observed variability in N and bromide was in general less than that
reported in other controlled field studies. Silvertooth et al. (1992) examined N and
bromide movement in an irrigated cotton production system in southern Arizona.
They indicate that their data "reinforce the presence of substantial variability
associated with the measurements of both nitrate and bromide at any date or depth of
sampling." CV values and SE^/X ratios calculated from their reported mean nitrate
concentrations (mg kg"1) vary from 0.12 to 1.39, and 12 to 40% through time,
respectively.
Table 5 shows that the CV values observed for mean N solute mass in
the soil profiles in the current study display a narrower range, from 0.58 to 0.99.
The SEx/X ratios observed here were substantially less, 7 to 12%, than those by
Silvertooth et al. (1992). The corresponding SE^/X ratios for N unrecoverable (%) in
the current study also display narrow ranges: 4 to 17% to 60 cm depth; and 10 to
20% to 90 cm depth (Table 6). However, irrigated conditions present in the
Silvertooth et al. (1992) study are more conducive to non-uniform flow. Irrigation
might well increase variability compared to rain-fed conditions due to preferential
flow of solutes under saturated or near-saturated conditions during and/or after
irrigation.
71
In addition, CV values calculated from the mean bromide concentration data,
(mg kg'1) reported by Silvertooth et al. (1992) varied from 0.25 to 2.24, while SEj/X
ratios varied from 16 to 49% through time. In another study, Agus and Cassel (1992)
examined field-scale bromide transport under three tillage treatments in the Atlantic
Coastal Plain of North Carolina. The CV values and SEjj/X ratios calculated from
their reported mean bromide concentration (mg kg'1) data varied from 0.33 to 2.39
and 16 to 18%, respectively, under the three tillage treatments on one sampling date.
Table 8 indicates that the CV values for mean bromide solute mass (mg kg'1) varied
from 0.55 to 2.16 in this field. The corresponding SE%/X ratios varied from 9% to a
high of 26%. Both CV and SEx/X ratio values increased as time elapsed from the KBr applications in S92 and F93. The highest values occurred in F93 just prior to
the second surface application of KBr. This increase over time in spatial variability
indicates that movement differences are expressed slowly. Thus, the observed SE^/X
ratios for mean bromide solute mass in this study are not only substantially lower than
those reported by Silvertooth et al. (1992), but in three of the five time periods, the
ratios are also much less than those observed by Agus and Cassel (1992). The CV
values for bromide in this study display narrower ranges than those in either of these
1992 studies.
Silvertooth et al. (1992) also calculated mass balance estimates for bromide
recovery (%) across five sampling dates. The SEg/X ratios calculated from their
reported data vary from 8 to 32%, with all but one value being greater than 10%. In
contrast, the SE^/X ratios calculated for bromide unrecoverable (%) in the current
72
study are substantially lower and varied from only 2 to 4% to 60 cm depth and 2 to
7% to 90 cm depth across the sampling periods. The CV values from their study
ranged from 0.20 to 0.78 as compared to 0.17 - 0.31 to 60 cm and 0.14 - 0.47 to 90
cm in the current study.
In 1985, Bruce et al. examined the redistribution of bromide by rainfall
infiltration. They concluded that observed differences in bromide transport between
soil profiles in a drainageway and those occurring on slopes at various sites on a
sandy loam landscape were due to differences in soil characteristics and rainfall
infiltration. The Bt horizons of soil profiles located in the drainageway not only had
higher hydraulic conductivities than slope sites, but they also had significantly smaller
maximum clay contents which occurred approximately 0.5 m deeper than in slope
sites. They state that "these and other differences in physical characteristics of the
soil profiles in addition to effect of landscape position produced the observed
differences in bromide movement among sites" (Bruce et al., 1985). Although the
soil and terrain characteristics do vary across the landscape in this farm field (Tables
44 and 45, Appendix C), there appears to be very little hydrologic variability as
expressed in measured water content (Table 15).
Bathke et al. (1992) reported that solute transport varied across landscape
positions due to changes in soil profile characteristics.
Bromide moved deeper in
profiles with lower clay contents and higher hydraulic conductivities. They state that
lateral bromide movement "was greatest at headslope sites where water drainage
pathways have developed over a long period of time" due to a "decrease in
73
macroporosity and increase in clay content at the residual soil material boundary in
the B horizon" (Bathke et al., 1992) Solute leaching was only indirectly analyzed by
classes of landscape position in the current study via computed terrain attributes.
These comparisons provide strong evidence that the spatial variability in N and
bromide solute concentrations observed in this field which experiences a drier climate
is substantially less than observed in other controlled field studies. Additionally, the
increase in spatial variability over time for bromide indicates that movement
differences are expressed slowly in this field.
Multiple Regression Analysis
The Pearson’s correlation matrices for pairwise comparison of the N and
bromide leaching indices with the independent variables indicated that no single
variable consistently explained a substantial portion of the variability in leaching as
measured by any leaching index. Multiple regression analysis was employed to
determine the combined influence of these variables. The stepwise regression results
included continuous variables summarized in Tables 13-16 and one indicator variable
delineating two different soil profile types. This indicator variable, limedepI, was set
to a value of I if depth to CaCO3 < 100 cm and alternatively set to zero if depth to
CaCO3 > 100 cm. Interaction variables were then defined as the product of each
environmental variable and limedep I. This provided a way to distinguish whether the
depth to CaCO3had any effect on leaching of N and bromide. If a soil, terrain, or
water attribute is present in a regression equation as a simple variable, this indicates
that the attribute contributes to the model for soil profiles where the depth to CaCO3
74
is greater than 100 cm. If an interaction variable appears in a regression equation, it
would indicate that the attribute is important and contributes to the model only for
those profiles where the depth to CaCO3 < 100 cm.
Three approaches to prediction of leaching were attempted for each solute,
nitrate-N and bromide, during each time period (where appropriate) as outlined in
Table 19. The first approach examined whether the solute mass or critical depth level
measured at the end of the current time period could be explained as a function of not
only soil, terrain, soil water, and yield variables effective during that time period, but
also by the past solute mass and critical depth levels from the previous sampling date.
The second approach examined whether, alternatively, the increase or decrease in the
solute mass or critical depth from one sampling date to the next could be explained as
a function of the soil, terrain, soil water, and yield variables effective during that time
period. This second approach was utilized because N dynamics in the soil are
complicated, and perhaps examining the change in solute mass over time would be
more instructive relative to the amount of leaching or N movement than trying to
predict levels of N solute mass at a particular point in time. This approach was only
used for time periods in which no new N or bromide was applied. Lastly, the third
approach attempted to determine the predictive capabilities of effective soil, terrain,
and soil water variables with respect to the amount of leaching (% unrecoverable)
which occurred over the current time period as estimated by the mass balance
calculations. YieM variables were not included in the stepwise regression model
statements for this third approach, because plant uptake (related to yield) was
75
previously accounted for in the mass balance calculations. Tables 20 and 21 contain
abbreviations for the soil water, soil, plant, terrain variables used in multiple
regressions.
Table 19. Regression approaches to prediction of field-scale leaching.
Independent Variables
Dependent
Leaching Index
Variable
Plant
Uptake
& Yield1
SM and CD
from previous
time period
Soil
Terrain
Soil
Water
Solute Mass
✓
✓
✓
✓
✓
Critical Depth
✓
✓
✓
✓
✓
SM-Difference
✓
✓
✓
✓
CD-Difference
✓
✓
✓
✓
Balance - 60 cm
✓
✓
✓
Balance - 90 cm
✓
✓
✓
1 Included only when applicable (i.e., Fall 1992).
The critical depth at 80%, rather than at 25 or 50%, was chosen for inclusion
in the multiple regression analysis because: I) for bromide, the almost linear increase
in critical depth was nearly the same for all three levels calculated (Figure 12); and 2)
for N, the changes in critical depth over time nearly paralleled one another at each of
the three levels (Figure 8).
Table 20. Abbreviations for soil water variables.
Depth
1992
1993
1994
25 cm
TW2592
TW2593
TW2594
100 cm
TW10092
TW10093
TW10094
150 cm
TW15092
TW15093
TW15094
76
The baseline soil potassium and phosphorous values listed in Table 43
(Appendix C) and summarized in Table 13 were not included as edaphic controls, or
variables, in the multiple regressions. Values for these nutrient levels were not
available for any time period after baseline values were obtained from initial soil
samples. N-P-K fertilizer was applied to the field in both S92 and F93 after soil
sampling as indicated in Table 3 and potassium and phosphorous levels were not
subsequently re-tested after either addition. Aspect was also not utilized as a variable
in the multiple regressions. Although aspect has a significant effect on
evapotranspiration and soil water content, the sites in this field exhibit mostly
southerly aspects.
Each approach was executed on a data set which included all sites (Stream
Channel In = SCI; n = 67) and on a smaller subset which excluded the sites lying
within the ephemeral stream channel (Stream Channel Out = SCO; n = 59). As
discussed earlier, the ephemeral stream sites receive additional water from outside the
study area via a culvert located on the east side of the field (Figure 2). Several of
v
these stream sites have higher specific catchment areas (SCA) (Table 44, Appendix
C), and thus wetness indices (WI), due to this outside influence which are not
quantifiable, because a DEM was not created for any area outside of the study area
boundaries. In other words, the SCA and WI values for these sites are biased
(underestimated) even though they are already higher than the values computed for
other sample locations. The sites considered to be in the stream channel were 30, 31,
34, 42, 43, 47, 55, and 69 (Figure.5).
77
’
Table 21. Abbreviations for soil, plant, and terrain attributes.
Attribute
Abbreviation
Soil
pH
pH
Electrical Conductivity
EC
Organic Matter
OM
Thickness of Mollic Epipedon
Mollic
Plant - Fall 1992
Nitrogen
Pl-Nit
Bromide
Pl-Br
Grain Yield - hand sampled
GY-Hand
Grain Yield - plot combine
GY-Avg
Terrain
Elevation
Elev
Slope
Slope
Specific Catchment Area
SCA
Wetness Index
WI
Plan Curvature
PlanC
Profile Curvature
ProC
Nitrogen. Table 22 contains the multiple regression results for N solute mass
present in the soil profiles at the beginning of the field monitoring period in S92.
The significant terms are listed in step order from Proc Reg in SAS which lists the
terms in order of decreasing partial
I2
values. These are the variables that were each
statistically significant at greater than the 95 % probability level and explained the
greatest amount of variation in N solute mass. Interaction variables are marked with
78
an asterisk. The results were dependent upon inclusion (SCI) or exclusion (SCO) of
the sites located in the ephemeral stream channel.
Table 22. Multiple regression results for nitrogen solute mass in Spring 1992.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out fn = 59)
Intercept
pH
7276.7
230.3
WI
pH I*
-68.4
SCA I*
PlanC I*
Elev
PlanC
(WI)
3.3
-27.3
-5.4
13.9
-8.8
Slope
0.20
0.25
0.20
14.56
0.0003
0.46
26.36
0.07
0.07
0.03
0.04
0.03
(0.01)
0.02
0.53
0.60
0.63
0.67
0.70
0.68
0.71
8.33
8.86
4.82
6.25
4.57
1.92
0.0001
0.0055
4.10
0.0043
0.0325
0.0156
0.0373
0.1717
0.0480
Stream Channel In (n = 67)
Intercept
Mollic
Elev
EC
SCA I*
Mollic I*
9607.4
6.2
-6.4
2211.5
3.5
-8.2
0.26
0.09
0.07
0.05
0.07
0.26
0.35
0.42
0.47
0.55
22.44
9.21
7.66
6.32
9.74
PlanC I*
-16.8
0.03
0.58
4.59
0.0001
0.0035
0.0074
0.0146
0.0028
0.0361
* Interaction Variable: variable * limedepl
The first column indicates that the model for SCO included seven terms and
the third and fourth columns indicate that the variability in these terms combined to
explain 71% of the variability in N solute mass. The wetness index (WI) was
79
removed from the model after the addition of plan curvature (PlanC). Pairwise
correlation shows that these two variables are related by an r-value of 0.66. For SCI,
the number of significant terms dropped to six and the variability in these terms
combined to explain 58% of the variability in N solute mass.
The derivation of the response functions that apply to different types of sites
may help illustrate the significance and meaning of the regression terms further. The
basic equation for a first order regression model with one interaction term is:
Yj = $o "b /^lxI "b f
e
where Yi is the leaching index,
+
0 3 X 1X2
ei
(7)
X1
might represent specific catchment area (SCA), x2,
might represent limedepl (whose value may be 0 or I depending on the location of
CaCQ3 in the soil profile), /30 is the intercept, /S1,
j6
2,
and /S3 represent the parameter
estimates for these two terms and their interaction (ie.
X1X2)
and
Ci
represents error.
The resulting response function would read:
E(Y) = 00 +
/J 1X1
+
/J2X2 + /J3X1X2
(8)
The exact meaning of the significant regression terms is best interpreted by
examining the nature of the response function for each lime unit or profile type
separately. The regression results for SCO summarized in Table 22 can be written in
equation form as:
E(Y) = 7276.7 + 230.Sx1 - 68.4xjX6 + 3.3x2x6 - 27.3x3x6 - 5.4x4
+ '13.9 x3 - 8 . 8 x 5
(9)
80
where Y is N solute mass in the soil profile,
is elevation, x5 is slope, and
X6
X1
is the pH,
X2
is SCA, x3 is PlanC,
X4
is limedepl.
Since limedepl is equal to I for sites where CaCO3 < 100 cm from the soil
surface, the x6 is equal to I, and the response function reads:
E(Y) = 7276.7 + 162.2X, + 3.3x2x6 - 13.4x3 - 5.4x4 - 8.Sx5
(10)
E(Y) = 7276.7 + 162.2(pH) + 3.3(SCA) - 13.4(PlanC) - 5.4(Elev)
- 8.S(Slope)
(11)
or:
indicating that predicted N solute mass in the soil profile (to a depth of 180 cm)
increases with increasing pH and SCA and decreasing PlanC, elevation, and slope.
The same general pattern is true for other types of sites where CaCO3 > 100
cm from the soil surface, although the absence of interaction terms changes the final
response function. For these sites, x6 (limedepl) is equal to zero, so that the response
function reduces to:
E(Y) = 7276.7 + 230.3x, + 13.9x3 - SAx4 - 8.8x5
(12)
E(Y) = 7276.7 + 230.3(pH) = 13.9(PlanC) - 5.4(Elev) - 8.S(Slope)
(13)
or:
i.
where predicted N solute mass increases with increasing pH and PlanC and decreasing
elevation and slope.
81
The regression models for SCO and SCI in Table 22 explained 58 to 71% of
the variation in N solute mass in the soil profile at the beginning of the field
monitoring period in S92. Twenty-nine to 42% of the variation remains unexplained.
No model was generated (SCO or SCI) for the critical depth of the N solute
mass in S92. In F92, prediction of N solute mass, critical depth, or changes in these
'
'
\
indices was not attempted. Nitrogen fertilizer was added to the field after the S92
sampling and, therefore, cannot be accounted for in a model. However, the mass
balance indices incorporate the N fertilizer addition. Although no model was
generated for the 90 cm depth, the results for N balance'to 60 cm in F92 are shown
in Table 23. The model for SCO (R2 = 0.17) indicates that % N unrecoverable
increases with decreasing pH and SCA. Sites with low SCA typically occur on
hilltops and near the tops of slopes. The model for SCI (R2 = 0.32) indicates that %
N unrecoverable increases at sites with lower SCA and more organic matter (OM).
An additional term appears in the model for soil profiles where CaCO3 < 100 cm;
the % N unrecoverable also increases as average total soil water to 150 cm depth
(TW15092) increases.
In S93, the N solute mass and mass balance indices produced the strongest
relationships (Table 24). The SCO model (R2 = 0.50) indicates that N solute mass
increases with increasing values of SCA, depth of the mollic horizon (Mollic), and
prior amount of N solute mass present in the profile in Fall 92 (NF92SM). The
I
model for SCI (R2 = 0.36) indicates that N solute mass increases with increasing
values of Mollic, NF92SM, and TW15092.
82
Table 23. Multiple regression results for nitrogen balance to 60 cm in Fall 1992.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out (n = 57)
Intercept
pH
SCA
97.4
-6.1
-0.01
0.10
0.07
0.10
0.17
5.93
4.44
0.0182
0.0398
0.13
9.07
0.0038
0.25
0.32
9.65
6.13
0.0029
0.0162
Stream Channel In (n = 60)
Intercept
SCA
31.8
0.13
-0.003
0.12
6.5
OM
0.07
TW15092 I*
0.2
* Interaction variable: variable * limedepl
Table 24. Multiple regression results for nitrogen solute mass in Spring 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out (n = 57)
Intercept
SCA
Mollic
NF92SM
56.0
0.2
2.0
0.6
0.32
0.12
0.32
0.44
0.06
0.50
25.54
11.19
6.73
0.0001
0.0015
0.0122
18.54
6.52
5.55
0.0001
0.0131
0.0217
Stream Channel In (n = 65)
Intercept
Mollic
NF92SM
TWl 5092
-111.8
2.4
0.5
5.0
0.23
0.07
0.06
0.23
0.30
0.36
NF92SM = N solute mass in soil profile in Fall 92.
83
The N mass balance results for S93 are listed in Table 25. In the top 60 cm,
the SCO model (R2 = 0.47) indicates that % N unrecoverable increases as SCA
decreases and PlanC increases. No model was generated to 60 cm for SCI because
none of the independent variables were significant. The SCO model to 90 cm (R2 =
0.55) indicates that % N unrecoverable increases as SCA and pH decrease and PlanC
increases. One additional term appears in the model for sites where CaCO3 < 100
cm; the % N unrecoverable increases as Mollic increases. The model generated for
SCI to 90 cm (R2 = 0.14) is only applicable to sites where CaCO3 < 100 cm. At
these sites, the model indicates that % N unrecoverable increases at sites with higher
wetness index (WI) values.
In F93, the models yielding the largest R2 values were those for N solute mass
and critical depth. Table 26 shows that the SCO model for N solute mass (R2 =
0.24) contains only one term, the prior amount of N solute mass present in the soil
profile in S93 (NS93SM). More N solute mass was present at sites which had a
higher amount of N solute mass in S93. The SCI model (R2 = 0.41) contains three
different terms. The NS93SM term was the first term entered in the model, but
subsequently removed from the final model. This model indicates larger amounts of
N solute mass at sites which have thicker mollic horizons, more water present to 150
cm, and are located at lower elevations.
84
Table 25. Multiple regression results for nitrogen balance in Spring 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out
60 cm (n = 51)
Intercept
SCA
PlanC
84.3
-0.03
0.8
0.29
0.18
0.29
0.47
19.80
16.17
0.0001
0.0002
131.6
-0.03
0.24
0.24
10.24
-9.7
0.16
0.40
8.44
0.0031
0.0067
0.8
0.08
0.48
4.73
0.0376
0.2
0.07
0.55
4.29
0.0472
6.02
0.0191
90 cm (n = 34)
Intercept
SCA
PH
PlanC
Mollic I*
Stream Channel In
60 cm (n = 58) - NOT SIGNIFICANT
90 cm (n = 38)
Intercept
WI I*
61.6
2.2
0.14
0.14
* Interaction Variable: variable * limedepl
The SCO model for N critical depth (R2 = 0.16) indicates that as slope
decreased, the critical depth of the N solute mass increased or moved deeper (Table
27). The SCI model (R2 = 0.34) indicates that N critical depth increased at sites with
more water present to 150 cm depth and lower pH values. An additional term
appears in the model for sites where CaCO3 < 100 cm. At these sites, N critical
depth increases as PlanC decreases.
85
Table 26. Multiple Regression results for nitrogen solute mass in Fall 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out (n = 56)
Intercept
NS93SM
247.0
0.8
0.24
0.24
17.23
0.0001
20.64
8.51
4.14
3.90
6.62
0.0001
0.0049
0.0462
0.0529
0.0126
Stream Channel In (n = 64)
Intercept
NS93SM
Mollic
TW15093
(NS93SM)
Elev
7990.3
4.6
10.9
-5.4
0.25
0.09
0.04
(0.04)
0.06
0.25
0.34
0.38
0.34
0.41
NS93SM = N Solute Mass in soil profile in Spring 1993.
Table 27. Multiple regression results for nitrogen critical depth in Fall 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out (n = 56)
Intercept
Slope
128.1
-2.2
0.16
0.16
10.56
0.0020
16.46
7.32
4.42
0.0001
0.0088
Stream Channel In (n = 64)
Intercept
TW15093
PlanC I*
113.4
1.7
-2.2
PH
-13.1
0.21
0.21
0.08
0.05
0.29
0.34
0.0398
* Interaction Variable: variable * limedepl
/
86 ^
The three models generated for % N unrecoverable in F93 (Table 50,
Appendix D), contained only one term for both SCO (R2 = 0.11) and SCI
(R2 = 0.08). The models indicate that % N. unrecoverable increases as SCA
decreases.
In S94, prediction of N solute mass, critical depth, or changes in these indices
was not attempted. Nitrogen fertilizer was again added to the field after the F93
sampling and cannot be accounted for in a model. The N mass balance model to 60
cm for SCO (R2 = 0.27) indicates that % N unrecoverable increases as OM content
and electrical conductivity (EC) decrease (Table 28). For sites where CaCO3 < 100
cm, one more term is added to the model. At these sites, % N unrecoverable also
increases as SCA decreases. The SCI model to 60 cm (R2 = 0.18) shows that % N.'
unrecoverable increases with decreasing ProC. Another term is again added for sites
where CaCO3 < 100 cm such that % N unrecoverable increases as EC decreases.
The models for SCO ( R2 = 0.23) and SCI (R2 = 0.25) to 90 cm contain the
same terms, but each term accounts for differing amounts of variability in each model
(Table 28). Both models indicate the % N unrecoverable increases as ProC
decreases. For sites where CaCO3 < 100 cm, % N unrecoverable is larger where
SCA is smaller.
Crop yield or plant uptake factors were not available for F94 and therefore, N
balance indices were not computed for this time period. However, prediction of N
solute mass and critical depth was attempted without values for these factors. The
models generated for N solute mass were identical for SCO and SCI (R2 = 0.27).
The models indicate that N solute mass increased at sites that contained more N solute
87
mass in the prior S94 sampling and have thinner mollic epipedons. No model was
generated for the change in solute mass from S94 for F94.
Table 28. Multiple regression results for nitrogen balance in Spring 1994.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out
60 cm (n — 49)
Intercept
SCA I *
OM
EC
79.4
-O l
-6.5
-129.1
O il
5.66
4.61
4.70
0.0214
0.08
0.07
O il
0.19
0.27
0.13
0.09
0.13
0.23
6.98
5.24
0.0113
0.0269
0.0371
0.0347
90 cm (n = 47)
Intercept
SCA I*
ProC
42.9
-O l
-10.6
.
Stream Channel In
60 cm (n = 56)
Intercept
50.3
ProC
-16.1
- 0.09
0.09
5.51
EC I*__________ -133.1________ 009________ 018________ 5.61
0.0226
0.0216
90 cm (n = 52)
Intercept
ProC
42.5
-14.8
0.15
0.15
8.81
0.0046
SCA I*___________ :0 1 ________ OlO________ 025________ 6.86
0.0117
* Interaction Variable: variable * limedepl
The models generated for F94 N critical depth (Table 29) and the change in N
critical depth (Table 30) were the same for SCO and SCI with the exception of model
R2 apd partial r2 values. The SCI models yielded slightly higher model R2 values in
88
each case. Table 29 indicates that N critical depth increased at sites with thinner
mollic epipedons and lower slopes. Table 30 indicates that the change in N critical
depth was greater (N moved deeper in profile) at sites with thicker mollic epipedons,
more OM, and lower pH values.
Table 29. Multiple regression results for nitrogen critical depth in Fall 1994.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out (n = 52)
Intercept
Mollic
Slooe
164.4
-0.7
-3.1
0.16
0.14
0.16
9.52
0.30
10.11
0.0033
0.0026
21.02
8.34
0.0001
0.0056
Stream Channel In (n = 57)
Intercept
Mollic
Slooe
165.2
-0.9
-2.6
0.28
0.10
0.28
0.38
Table 30. Multiple regression results for change in nitrogen critical depth in Fall
1994.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out (n = 52)
Intercept
OM
-194.4
21.8
0.12
0.12
6.66
0.0128
Mollic
32.5
0.14
0.26
9.26
0.0038
dH
-0.8
0.08
0.34
5.91
0.0188
8.90
11.78
5.84
0.0043
0.0012
0.0192
Stream Channel In (n = 57)
Intercept
Mollic
OM
-JlH________
-176.2
20.1
29.8
-1.0
0.14
0.15
0.07
0.14
0.29
0.36
89
Bromide. Although bromide (KBr) was added to the field after the S92
sampling, prediction of the change in bromide solute mass and critical depth indices
was attempted. The changes were calculated as the differences from zero since no
bromide was present in the soil profiles prior to the S92. Therefore, these differences
reflect the amount of bromide present in the profiles in F92. Table 31 shows that the
models for change is bromide solute mass are very similar for SCO (R2 = 0.29) and
SCI (R2 — 0.31). The models are only applicable for sites where CaCO3 < 100 cm.
These results indicate that sites with smaller PlanC values and more water present in
K
the profiles experienced greater changes in bromide solute mass. In other words,
more bromide remained in the soil profile, therefore, less bromide leaching occurred
at these sites.
Table 3 1. Multiple regression results for change in bromide solute mass
from Spring 1992-Fall 1992.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out (n = 57)
Intercept
PlanC I*
TW2592 I*
137.0
-10.5
13.5
0.19
0.10
0.19
0.29
13.00
8.03
0.0007
0.0065
16.31
9.89
0.0001
0.0026
Stream Channel In (n = 65)
Intercept
TW15092 I*
PlanC I*
134.0
2.5
-9.8
0.21
0.21
0.11
0.31
* Interaction variable: variable * limedepl
90
The F92 bromide mass balance indices do incorporate the bromide application
in S92 (Table 32). All four models generated for % bromide unrecoverable to 60 and
90 cm were very similar for both SCO (R2 = 0.15 - 0.16) and SCI (R2 = 0.16 0.24) and are applicable only for sites where CaCO3 < 100 cm. Each model
indicates a positive relationship between % bromide unrecoverable and PlanC. One
additional term is incorporated into the model for SCI to 90 cm. This model indicates
an inverse relationship with EC.
Table 32. Multiple regression results for bromide balance in Fall 1992.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out
60 cm (n = 53)
Intercept.
PlanC I*
54.1
1.4
0.16
0.16
10.07
0.0026
40.1
1.3
0.15
0.15
6.70
0.0214
90 cm (n = 40)
Intercept
PlanC I*
Stream Channel In
60 cm (n = 60)
Intercept
PlanC I*
54.8
1.4
0.16
0.16
11.46
0.0013
0.16
0.08
0.16
7.95
0.24
4.51
0.0072
0.0397
90 cm (n = 45)
Intercept
PlanC I*
EC I*
46.1
1.1
-135.1
* Interaction Variable: variable * limedepl
91
The S93 results for bromide solute mass, critical depth, and the changes in
these indices were either poor or no models were generated at all (Tables 52 -5 4 ,
Appendix D). Table 33 list the models for bromide balance in S93. The models are
virtually identical for SCO (R2 = 0.36) and SCI (R2 = 0.31). Both indicate that %
bromide unrecoverable increases at higher elevations. Additionally, both models
indicate that for sites where CaCO3 < 100 cm, % bromide unrecoverable also
increases at sites with thinner mollic epipedons.
Table 33. Multiple regression results for bromide balance in Spring 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out
60 cm (n = 48)
Intercept
SCA
87.8
-0.01
0.08
0.08
4.27
0.0444
0.24
0.12
0.24
8.46
4.72
0.0073
0.0395
7.28
4.50
0.0115
0.0429
90 cm (n = 28)
Intercept
Mollic I*
Elev
-1256.0
-0.9
0.9
0.36
Stream Channel In
60 cm (n = 55) - NOT SIGNIFICANT
90 cm (n = 31)
Intercept
Mollic I*
Elev
-1162.8
-0.8
0.8
0.20
0.11
* Interaction Variable: variable * limedepl
0.20
0.31
92
In F93, no models were generated for the change in critical depth or %
bromide unrecoverable to either 60 or 90 cm. Models for bromide solute mass (R2 =
0.13 - 0.15) and the change in solute mass (R2 = 0.14 - 0.21) (Table 55, Appendix
D) were very similar for both SCO and SCI. Each model indicates a positive
relationship between the solute mass index and pH value. The models for bromide
critical depth yielded fairly high R2 values and were very similar for SCO and SCI.
The sample numbers are low because the regressions were run only for sites where
bromide was still present within the top 180 cm. In both cases, pH was the first
significant term added to the model, but was removed from the final model. For
SCO (R2 = 0.65), bromide critical depth increases as both ProC (convex) and EC
decrease (Table 34). For SCI (R2 = 0.72), one new term is added to the model such
that bromide critical depth is inversely related to not only ProC and EC, but also to
SCA.
As discussed earlier, a second treatment of bromide was surface applied in
&
'
F93 just prior to soil sampling. In S94, regression results for change in bromide
solute mass for SCO (R2 = 0.17) indicate greater differences (ie. more bromide still
present in profile; less leaching occurred) at sites with higher SCAT and lower PlanC
values (Table 57, Appendix D). No model was generated for SCI data set. The same
model (R2 = 0.13) was generated for change in critical depth for both SCO and SCI
(Table 58, Appendix D). This model shows that greater downward changes in critical
depth occur at sites with smaller ProC values.
93
Table 34. Multiple regression results for bromide critical depth in Fall 1993
(for sites where solute present at < 1 8 0 cm).
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out (n = 22)
Intercept
pH
ProC
EC
185.4
0.27
0.29
0.09
(0.004)
-37.6
-595.4
(DH)
0.27
0.56
0.65
0.65
7.43
12.45
4.87
0.18
0.0130
0.0022
0.0407
0.6723
Stream Channel In (n = 24)
Intercept
/
192.1
PH
ProC
EC
-33.7
-605.4
(pH)
SCA
-0.2
0.27
0.27
8.11
0.0094
0.28
0.10
(0.006)
0.07
0.55
0.65
0.64
0.72
12.93
5.94
0.16
5.19
0.0017
0.0289
0.6933
0.0338
The S94 model for bromide mass balance to 60 cm for SCO (R2 = 0.09)
indicates that % bromide unrecoverable increases at sites with higher slopes (Table
59, Appendix D). No model was generated for SCI. The model for % bromide
unrecoverable to 90 cm was the same for SCO and SCI with an R2 of 0.08. It shows
a positive relationship with EC.
Finally, in F94, no models were generated for bromide critical depth or
change in bromide solute mass. Table 35 lists the results for bromide solute mass in
F94. Once again, the models for SCO (R2 = 0.21) and SCI (R2 = 0.25) are almost
identical. Both models are applicable only for sites where CaCO3 < 100 cm and
indicate a positive relationship with OM content and an inverse relationship with
94
slope. The models for change in critical depth for F94 yielded the same R2 value .
(0.08) and indicated an inverse relationship with slope for SCO and an inverse
relationship with the thickness of the mollic epipedon for SCI (Table 60, Appendix
D).
Table 35. Multiple regression results for bromide solute mass Fall 1994.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
Stream Channel Out (n = 52)
Intercept
OM I*
342.1
111.7
Slooe I*
-20.1
0.13
0.07
0,13
0.21
7.77
4.68
0.0075
0.0354
12.73
4.29
0.0008
0.0431
Stream Channel In (n = 57)
298.7
Intercept
0.19
121.9
OM I*
0.06
-19.0
Slone I*
* Interaction variable: variable * limedepl
0.19
0.25
Discussion. Results for regressions not discussed are listed in Tables 45-60 in
Appendix D. Tables 22 - 35 discussed above (and those in Appendix D) illustrate
that the attempt to predict nitrate or bromide leaching occasionally showed some
promise, but consistency was a problem. Model R2 values obtained were typically
low and no consistent patterns or groupings of significant independent variables were
identified for either N or bromide. When model R2 values were high (i.e., F93
critical depth; Table 34), the relationship between the leaching index and significant
terms do not appear to make explicit sense or be physically reasonable. Although the
)
95
change in solute mass and critical depth indices provide the most direct method of
characterizing leaching, none of the three approaches to multiple regression analysis
produced results that were more consistent or stronger than the other two. No
consistent trends were observed in the presence of outliers on scatterplots. Based on
Mallow’s Cp values, some models are over-specified and some are under-specified
with respect to the number of significant terms included in the model. Nitrogen
appears to, on average, show stronger relationships to the soil, soil water, and terrain
variables than bromide. One reason for this trend may be related to N cycling and
mineralization.
Relationships between conventional soil morphology and edaphic properties
directly relevant to land management under irrigated or dryland systems are often
weak (McKenzie and MacLeod, 1989). This study, therefore, utilized a combination
of conventional soil morphological characteristics (thickness of mollic horizon and
depth to CaCO3) and soil properties measured analytically in the lab (i.e., pH, OM,
EC, etc.). Soil textures were not utilized in this study as one of the independent soil
variables. Clay percentages estimated from the initial soil cores removed from the
field were listed as 26% (with only three exceptions).
Bruce et al. (1985) state that variability in topography and soil characteristics
from place to place on the landscape affecting infiltration and soil water transport
frequently does not permit ready description of solute distribution in fields.
Differences in environmental attribute correlations and model development can occur
due to physiographic setting, scale of analysis, computation methods, and other
96
reasons (data structure, quality and error) (Gessler, et al., 1993). In this field, small
differences in leaching across a range of topographic sites were observed. In light of
the paucity of spatial variability of water and solute observed, prediction of leaching
differences is statistically difficult where variability across the landscape is negligible.
This level of variability might be below the minimum detection level for prediction.
It is probable that the overall results might not be different if another model
for solute leaching had been used. These results illustrate fundamental questions
associated with appropriate sampling in space and time. Soil-landscape processes
operate across a range of spatial and temporal scales (Kachanoski, 1988). This
complexity causes soil properties to exhibit different and complex scales of variation
(Burrough, 1993). Therefore, "expectations for deciphering the relationship between
pattern and process should vary in different physiographic domains" (Gessler et al.,
1993). The scale of the predicted variable must match scale of the process of
interest, ie. field-scale leaching, with respect to both time and space. If the chosen
scale is below the scale of the process, there may be too much "noise" in the data to
identify pertinent relationships. If the chosen scale is above the scale of the process,
it is possible to miss important information.
With respect to space, the coarse 10 m DEM grid used for generation of
terrain attributes was originally chosen for applicability in site-specific farming. Ten
meters may be the smallest operational unit manageable by a farmer or within which
equipment may be regulated on-the-go in a field (Nielsen, 1994). However, digital
terrain attributes are scale dependent and computed attributes may be meaningless or
97
the processes of interest may be masked if scale effects are not considered (Moore et
aL, 1991; 1993c).
'
■
)
.
Terrain attributes computed from digital terrain models can be measures of the
hydrologic processes based on surface form at the hillslope scale: I) water
redistribution through lateral flow; and 2) sediment transport through runoff. Neither
process may have occurred in this field to any significant degree during the duration
of the study. Alternatively, geomorphic attributes are measures of broader scale
regional processes. Aeolian deposition produces a progressive fining or thinning of
loess at greater distances away from mountains. The Spiinghill study site is located
near the base of the Bridger Mountains and possesses a thick loess cap. It is possible
. that the thick, permeable, loess-derived soils capture and retain precipitation from
rainfall and melting snow without significant lateral redistribution in the landscape.
No surface runoff, or evidence thereof* was observed in three field seasons, except at
the ephemeral stream sites in Spring 1994.
Water and solute transport in soils are intimately related to soil property and
water balance relationships (Bathke et al., 1992). Additionally, Afyuni et al. (1994)
state that the vertical and lateral anisotropic nature of soils plus any downslope
hydraulic gradient enhances lateral as well as vertical chemical transport. However,
Jury et al. (1982) showed in a field test that the distribution of relative water uptake
by wheat in the root zone had a greater effect on solute concentration distributions
than the soil water content.
I
98
Only very small differences in solute movement were observed across a range
of topographic positions and soil properties in this study. The results presented here
seemingly indicate no connection between soil or terrain variables and leaching in the
short term. It is possible that leaching in this environment is driven by pulses of
water or events that were not monitored with this sampling scheme. Neutron probe
data were by necessity never collected while it was raining, and usually around 48
hours later.
The water data indicate that water is lost quickly in this environment or
that most values measured were near field capacity, especially below the zone of
substantial root uptake. Additionally, it is possible that lateral water flow rarely
occurs in this field and that the time scale of observation of leaching in this study may
not have been sufficient to expose any real differences in solute movement relating to
water balance in the soil. The increase in the spatial variability of bromide over time
seems to support this in that it indicates that movement differences are expressed
slowly in this field.
Finally, the presence of gopher holes around many of the neutron access tubes
may have contributed to measurement error with respect to soil moisture readings and
solute concentrations (if soil core passed through underground tunnel).
Conclusions
An attempt was made to identify and quantify, by multiple stepwise regression,
the environmental factors that regulate water flow and leaching within a dryland
farming system in southwest Montana. The factors considered included edaphic,
topographic, soil water, and biotic characteristics. The observed variability in
99
measured water content across the soil landscape was small.
The data indicate that
nitrate and bromide leaching did occur. Due to the minimal lateral redistribution of
water in this field, removal or loss of nitrate and bromide not accounted for by plant
uptake are attributed to leaching. However, the observed variability in nitrate and
bromide leaching was also small and less than reported in other field studies.
Nitrate and bromide exhibited statistically significant differences in spatial
behavior patterns in all but a couple of time periods. The uniformity of the water and
solute data across a range of topographic positions made prediction of leaching
"differences" statistically difficult. Differences in leaching across the deep,
permeable soil landscape were not satisfactorily explained with the combination of
soil, water, and terrain attributes (10 m scale) in this farm field. Multiple regression
analyses occasionally showed some promise, but were not consistent. Overall, the
nitrate indices produced slightly stronger relationships to the environmental factors
than bromide.
These results reiterate the difficulties that are encountered in trying tb predict
(understand) the interrelationships between environment, N leaching, and potential
contamination of ground water. In order to prevent ground water contamination, it is
important not only to identify and evaluate the characteristics of soils and the
landscape critical to water and solute distribution, but also those which account for
the variability of observed field-scale leaching.
100
CHAPTER 3
RELATIONSHIP BETWEEN MEASURED SOIL WATER CONTENT
AND A TOPOGRAPHICALLY-DERIVED WETNESS INDEX
Introduction
Qualitative and quantitative models of soil-landscape relationships have been
developed by numerous researchers (Aandahl, 1948; Troeh, 1964; Walker et al.,
1968; Speight, 1974; Webster, 1977; Dikau, 1989; Odeh et al., 1991; Moore et al.,
1988b, 1993a; 1993b). Several researchers have asserted that topographic features
specifically influence surface runoff, subsurface water movement, the development of
zones of surface saturation, and the distribution of soil water content (Zaslavsky and
Rogowski, 1969; Anderson and Burt, 1978; Beven and Kirkby, 1979; Zaslavsky and
Sinai, 1981a-d; Sinai et al., 1981; O’Loughlin, 1981, 1986; Hanna et al., 1982,
1983; Burt and Butcher, 1985; Moore and Burch, 1986).
Most topographic attributes are defined at points within a catchment on the
landscape. However, maps of these landscape parameters or topographic attributes
are easily derived from a digital elevation model (DEM) within a GIS framework
(Moore et al., 1988a). In 1989, Dikau stated that the most important quantifiable
landform attributes which can be computed from a DEM are slope, plan curvature,
and profile curvature. However, Odeh et al. (1991) added upslope distance and area
101
attributes and reported that these five attributes accounted for the greatest majority of
the soil variability in their study area (Spangmd, 1995). Moore et al. (1993a) have
related soil variability to compound topographic attributes such as wetness, stream
power, and sediment transport capacity indices.
They have proposed to expand their
model to more heterogeneous landscapes through the inclusion of geology, climate,
and radiation variables (Spangmd, 1995).
Jones (1986; 1987) discusses the usefulness of the wetness index as an
indicator of the spatial distribution of soil water content and soil water drainage. He
states four assumptions or limitations involved in its use: I) a spatially uniform
infiltration capacity; 2) a spatially and temporally uniform transmissivity; 3) a
constant subsurface leakage to ground water; and 4) coincident surface and subsurface
drainage areas (Moore et al., 1988a). Burt and Butcher (1985) also state that "care
must be exercised in applying a static index, like the wetness index, to predict the
distribution of a dynamic process like soil water content that shows hysteric effects,
especially when threshold changes occur in the area of saturation" (Moore et al.,
1991).
Beven and Kirkby (1979) related the topographic structure, of a catchment and
the average soil water storage to the saturated area (Moore et al., 1988a).
O’Loughlin (1986) derived a slightly different form of wetness index that is capable
of handling spatially variable transmissivities. However, both authors relate their
wetness index to the spatial distribution and size of zones of saturation or variable
source areas for runoff generation (Moore et al., 1991). Moore and Nieber (1989)
102
suggest that "a threshold concept could possibly be used to also relate the potential for
ground water recharge and non-point source pollution to a wetness index because
significant percolation of soil water to ground water occurs only when the soil water
content exceeds field capacity."
Burt and Butcher (1985) found that the product of a wetness index and plan
curvature ,produced the best correlations with surface soil water distributions.
Alternatively, Moore et al. (1988a) indicate that a linear relationship involving a
wetness index and aspect yielded the best results. Moore et al. (1988b) also reported
a strong correlation between the distribution of a wetness index and the distribution of
surface soil water in a small fallow catchment.
Moore’s steady-state wetness index is an analytically-derived, compound
topographic attribute (Moore et al., 1991). It is calculated from two primary
attributes, specific catchment area and slope, and is described as:
w = ln(As/tan /3)
(14)
where w indicates the wetness index, As is the specific catchment area (m2m"1), and /3
is the slope angle (degrees). Two important assumptions are incorporated into this
wetness index: I) that the direction of subsurface flow (dictated by the gradient of
the piezometric head) is parallel to the gradient of the surface topography; and 2) that
steady-state conditions apply. The wetness index values are indices relative only to
the total area analyzed with the DEM. The value of this index increases with
increasing specific catchment area and decreasing slope gradient. Larger indices
103
indicate those cells in the study area having low slope and large drainage areas
relative to other cells in the study area. Therefore, hilltops (flat areas with low
specific catchment area) exhibit moderate values, valleys (high specific catchment area
and low slope) where water concentrates exhibit high values, and steep hillslopes
(high slope) where water drains more freely exhibit low values (Moore et al. ,
1993a; 1993b).
Moore et al. (1993a) hypothesized that "the spatial distribution of topographic
attributes that characterize water flow paths also captures the spatial variability of soil
attributes at the meso-scale" (IO4 to 2 * IO5 m). They state that the wetness index
has been used to characterize the spatial distribution of zones of surface saturation and
soil water content in landscapes (Moore et al., 1988a; 1993d) and appears to be useful
for mapping forest soils (Skidmore et al., 1991). Moore et al. (1993b) show that the
wetness index is correlated with several soil properties such as silt percentage
(r = 0.61), organic matter content (r = 0.57), phosphorous (r = 0.53), and A
horizon depth (r = 0.55) in the soil surface of a small toposequence in Colorado.
They reported that: I) thick A horizons on steep slopes are associated with high
values of wetness index and plan curvature (ie. concave areas where water
■
concentrates) and; 2) the wetness index appears to characterize similar patterns of
overland and subsurface flow convergence.
Gessler et al. (1993) found that the wetness index calculated from TAPES-G
was a useful predictor for some diagnostic morphologic properties because it
"combines contextual and site information via the upslope catchment area and slope
104
respectively." Bell et al. (1994) calculated a similar wetness index, defined as:
WI = log (A/S)
(15)
where A is the specific catchment area (m2 nr1) and S is the slope gradient (%).
They reported that this wetness index correlated with A-horizon depth (r = 0.57) and
depth to carbonates (r = 0.59) in a 20-ha study site located in west-central Minnesota
currently in a mixture of grasses under CRP (Bell et al., 1994).
Objectives
The objective of this study was to test whether Moore’s compound,
quantitative, topographically-derived wetness index calculated from a DEM at two
different grid scales (2 m and 10 m) is correlated to field-measured soil water content.
Materials and Methods
Study Site
The analysis was conducted on data from the study site described in Chapter 2
(p. 22): a 20-ha cultivated field of moderate relief (43 m) in southwestern Montana.
Depth of Mollic Horizon
The depth of the mollic horizon (cm) at each site was measured and recorded
from soil cores removed in Spring 1992 for installation of neutron access tubes
(Chapter 2).
105
Generation of Wetness Indices
The wetness indices at the 10 m scale were generated as described earlier in
Chapter 2 (p. 33) utilizing ANUDEM (Hutchinson, 1989) and the grid version of the
Terrain Analysis Programs for the Environmental Sciences (TAPES-G) (Moore,
1992). Slope was calculated using the finite difference approach and specific
catchment areas were calculated using the FRhoS algorithm. These procedures were
repeated using a 2 m regular rectangular grid to produce the second set of wetness
indices. The 10 m scale was originally chosen for its possible applicability to sitespecific farming (Nielsen, 1994). For this analysis, a scale of 2 m was chosen for
comparison as it more closely represents the physical sphere of influence of the
neutron probe in the field when taking measurements.
Soil Water Content
Collection of field-measured soil water content and calculations of average
total water in the profiles at the 70 sites is described in Chapter 2 (p. 36). For
purposes of this analysis, average total water content values to 25, 100, and 150 cm
in the soil profiles were also calculated for a sampling date in June of each year. The
June dates were June 22, June 25, and June 24 for 1992-1994, respectively. These
June dates were selected for comparison since they represent the "wettest" time of
year sampled, and therefore come closest to satisfying the assumption of steady-state
conditions in wetness index calculations in TAPES-G (Moore, 1992).
106
Statistical Analysis
As stated earlier in Chapter 2, three sample points (#52, #64, #65) were
removed from analysis due to clipping the DEM coverage to a rectangular area. The
wetness index, depth of mpllic horizon, and soil water variables were imported into a
spreadsheet for calculation of some summary statistics (CV, SEx, and SExZX). The
data were then exported to two ASCII files: one containing a full data set (n = 67)
referred to as Stream Channel In (SCI), the othef excluding the sites in the stream
channel (n = 59) referred to as Stream Channel Out (SCO). These files were
subsequently used as input files for statistical analysis. Additional summary statistics
were obtained from the SUMSTAT procedure in MSUSTAT (Lund, 1991).
Paired t-tests were performed between the season average water values and the
June water data using MSUSTAT. Correlation analyses were performed (MSUSTAT)
utilizing the NPCOER procedure to estimate Pearson’s correlation coefficients and Pvalues for pairwise comparison of the wetness indices (2 m and 10 m grids) with the.
soil water variables (season averages and June data) to 25, 100, and 150 cm depths.
Correlation analyses were also performed between the wetness indices and thickness
of the mollic epipedon. All correlations were calculated for both SCI and SCO. The
ephemeral stream sites removed for the SCO data set were 30, 31, 34, 42, 43, 47,
55, and 69. As discussed in Chapter 2, although these sites have wetness index
values higher than for other sites, they are biased (underestimated) due to the
influence of water imported from outside the study area and DEM boundaries.
107
Results and Discussion
Wetness Indices
Table 36 lists the summary statistics for the wetness indices. The minimum,
maximum, mean, standard deviation (SD), and coefficient of variation (CV) values
for the wetness indices are very similar at each scale for SCI. Although the minimum
values remain the same for SCO, the maximum values are lower at both scales.
There is also a greater difference in the standard deviation values between the two
scales for SCO. Table 61 in Appendix E contains a complete listing of the wetness
index values for each site at the 2 m and 10 m scales. These values are graphed in a
scatterplot Figure 13 for visual, comparison.
Table 36. Summary statistics for wetness indices.
Stream Channel In
(n = 67)
Stream Channel Out
(n = 59)
2m
10 m
2m
10 m
Min
3.833
4.533
3.833
4.533
Max
13.223
13.271
11.702
10.470
Mean
6.648
6.940
6.371
S.D.
1.769
1.771
1.486
1.132
C.V.
0.263
0.255
0.233
0.173
S.D. = standard deviation.
C.V. = coefficient of variation.
:
6.552
Figure 13. Scatterplot of wetness index values at 2 m and 10 m scales.
2 m Wetness Index
109
Table 37 lists the results of both Pearson’s correlation and Spearman rank
correlation analysis of the wetness indices. Both Figure 13 and Table 37 indicate that
the 2 m and 10 m wetness index values are only fairly well correlated to one another.
Pearson’s correlation analysis yielded an r-value of 0.74 for the SCI data set (R2 =
0.54; Figure 13). A slightly larger R2 of 0.71 was obtained for the SCO data set,
which removed three of the four outliers in Figure 13. Spearman’s rank correlation
analysis revealed an opposite trend. The value of r decreased from 0.81 for SCI to
0.79 for SCO. All r-values were significant at the 5 % level.
Table 37. Correlation analyses of wetness indices at 2 m and 10 m scales.
Correlation Procedure
Stream Channel In
(n = 67)
Stream Channel Out
(n = 59)
r
P-value
r
P-value
Pearson’s
0.7377
0.0000
0.8441
0.0000
Spearman Rank
0.8102
0.0000
0.7923
0.0000
Paired t-tests between the wetness indices were significant at the 10% level
(P = 0.067 for SCI; P = 0.091 for SCO) indicating that the 2 m and 10 m values do
not share similar patterns of spatial variability.
Soil Water Content
Table 38 lists the summary statistics for the average total water in the soil
profiles in the upper 25, 100, and 150 cm depths during the growing seasons from
1992 to 1994. As discussed in Chapter 2, the data reveal minimal differences in soil
water measured to the three depths across the landscape as evidenced by the small
coefficient of variation (CV) values (0.04 - 0.12) and SE^/X ratios (0.01 - 0.09).
no
Table 38. Summary statistics for season average total water (SCI).
25
cm
1992
100
cm
n
Min
67
5.64
65
14.41
Max
7.88
6.77
0.59
0.09
0.07
0.01
Mean
S .D .
C .V .
SEx1
SE x/X 2
150
cm
64
25
cm
1993
100
cm
25.50
18.24
25.90
55.98
65
6.79
9.39
32.78
8.36
61
22.56
29.19
26.02
1.71
3 .8 8
0.09
0.43
0.02
0.12
0.95
0.03
0.61
0.07
0.19
0.02
1.23
0.05
0.93
0.04
150
cm
25
cm
57
23.35
40.00
65
5.52
8.14
36.89
2.81
0.08
1.65
0.05
6.55
0.54
0.08
0.18
0.03
1994
100
cm
150
cm
61
19.14
43
29.07
25.15
34.58
32.42
21.49
1.17
0.05
0.77
0.04
1.31
0.04
1.92
0.09
1 Standard error of the mean.
2 Standard error of the mean as proportion of the mean.
This lack of spatial variability is also evidenced in the summary statistics for
total soil water present to the three depths on a sampling date in June of each year
(Table 39). The CV values are again small and exhibit a narrower range from 0.05
to 0.09.
As discussed in Chapter 2 (pp. 67-69), both the Pearson’s and Spearman Rank
correlation coefficients between the season average and June data were fairly high
regardless of inclusion or exclusion of ephemeral stream sites. Paired t-tests between
the season average water data and the June data yielded P values of 0.0000 for each
combination for both SCI and SCO; with one exception. The spatial pattern of season
average water data was most similar to the pattern of June data to 150 cm in 1993
(P = 0.50 for SCI and P = 0.92 for SCO).
J
Ill
Table 39. Summary statistics for total water in June (SCI).
1992 - June 22_______ 1993 - June 25_________1994 - June 24
25
cm
100
cm
150
cm
25
cm
100
cm
150
cm
25
cm
100
cm
150
cm
n
Min
Max
Mean
64
8.79
11.30
56
35.62
50.21
64
6.67
9.44
9.99
65
22.31
35.30
31.48
44.98
50
30.36
40.39
37.45
65
5.79
10.06
7.54.
58
20.49
27.75
24.27
42
28.89
37.88
35.24
S.D.
C.V.
0.66
0.07
1.76
0.06
2.32
0.05
8.18
0.62
61
21.18
29.03
25.77
1.31
0.05
1.96
0.05
0.71
1.29
0.05
1.84
0.08
0.09
0.05
Correlation Analysis
Pearson’s correlation matrices for the soil water variables and the 2 m and 10
m wetness indices are shown in Tables 40 (SCI) and 41 (SCO). From these matrices
it can be seen that some correlation coefficients were positive, some were negative.
Overall, the r-values were very low indicating that the topographically-generated
wetness indices do not relate well to field-measured water content (season average or
June date).
The correlations for the season average total soil water variables reveal no
pattern with respect to positive or negative values for certain depths for either the 2 m
or 10 m wetness index. The range of r-values for SCI were very similar for the 2 m
(| r | = 0.01 - 0.16) and 10 m ( | r I = 0.03 - 0.17) scales (Table 40). Removing the
sites in the ephemeral stream channel (SCO) did not substantially improve the
correlations. At 2 m, |r | = 0.02 - 0.19; and for the 10 m index, |r | = 0.02 - 0.21
(Table 41).
-
112
Table 40. Pearson’s correlation analyses between wetness indices and soil water
variables (SCI).
Total Soil
Water Attribute
n
2m
Wetness Index
10 m
Wetness Index
r
P-value
r
P-value
1992 Season Averaee
25 cm
67
-0.0604
0.6275
-0.0292
0.8145
100 cm
65
0.0899
0.4761
0.1734
0.1673
150 cm
64
-0.1041
0.4130
-0.1416
0.2642
25 cm
65
-0.0681
0.5897
-0.1042
0.4087
100 cm
61
-0.0254
0.8459
-0.0789
0.5454
150 cm
57
0.0653
0.9615
0.1166
0.3878
25 cm
65
-0.0155
0.9023
-0.0755
0.5502
100 cm
61
-0.0109
0.9334
0.0311
0.8119
150 cm
43
0.1595
0.3070
-0.0351
0.8232
25 cm
64
-0.0280
0.8259
-0.1002
0.4307
100 cm
65
0.0941
0.4560
0.0369
0.7705
150 cm
56
-0.0815
0.5505
-0.1436
0.2910
64
-0.0780
0.5403
-0.0910
0.4745
100 cm
. 61
-0.2350
0.0683
-0.2605
0.0426
150 cm
50
-0.0980
0.4986
-0.1400
0.3320
25 cm
65
0.2993
0.0154
0.1499
0.0154
100 cm
58
0.1844
0.1659
0.0815
0.5432
150 cm
42
0.2394
0.1268
0.0758
0.6333
1993 Season Average
1994 Season Average
1992 - June 22
1993 - June 25
25 cm
1994 - June 24
.
113
I
Table 41. Pearson’s correlation analyses between wetness indices and soil water
variables (SCO).
Total Soil
Water Attribute
2m
Wetness Index
n
r
10 m
Wetness Index
r
P-value
P-value
1992 Season Average
•
25 cm
59
-0.0434
0.7439
-0.0957
0.4711
100 cm
58
0.0798
0.5514
0.0250
0.8521
150 cm
58
-0.0160
0.9049
-0.0441
0.7421
25 cm
57
0.0766
0.5714
0.0557
0.6808
100 cm
55
0.1507
0.2720
. 0.1216
0.3766
150 cm
53
0.1321
0.3457
0.1943
0.1633
1993 Season Average
1994 Season Average
, t
‘
25 cm
57
0.0517
0.7026
0.0357
0.7918
100 cm
54
0.1833
0.1846
0.2084
0.1305
150 cm
41
0.1905
0.2329
0.0219
0.8916
25 cm
57
0.0792
0.5584
0.0426
0.9749
100 cm
58
0.1412
0.2904
0.1222
0.3610
150 cm
53
0.0810
0.5644
-0.0191
0.8919
25 cm
56
0.0427
0.7544
-0.0487
0.9716
100 cm
54
0.1604
0.2465
0.1042
0.4535
150 cm
47
0.1164
0.4360
-0.0831
0.9558
0.1988
0.1382
0.1538
0.2533
1992 - June 22
1993 - June 25
1994 - June 24
.
25 cm
57
100 cm
53
0.2975
0.0305
0.2097
0.1318
. 150 cm
40
0.2399
0.1359
0.0785
0.6300
■
114
The second approach involving total soil water present on June sampling dates
did yield a few correlations slightly higher than for the season average soil water
variables. The range of r-values for the SCI data set were |r | = 0.03 - 0.30 for the
2 m wetness index and |r | = 0.04 - 0.26 for the 10 m index (Table 40). Removing
the sites in the ephemeral stream channel (SCO) yielded the following correlations:
At 2 m, ]r| = 0.04 - 0.30; and for the 10 m index, |r | = 0.02 - 0.21 (Table 41).
Table 42 shows, that the wetness indices moderately correlate with the
thickness of the mollic horizon for both SCI ( |r | = 0.42 - 0.62) and SCO ( |r | =
0.36 - 0.41).
All r-values were significant at the 5% level. Thicker mollic horizons
tend to develop in the landscape at locations where either: I) large scale events have
historically placed more water; or 2) more water is present more often due to
microtopographic effects.
However, correlation analyses between soil water
variables and the thickness of the mollic horizon yielded low r-values of |r | =
0.03 - 0.23 for SCI and |r | = 0.01 - 0.29 for SCO (not shown). Additionally, most
soil water variables exhibited and inverse relationship to the thickness of the mollic
horizon.
Conclusions
Data from this cultivated farm field in southwestern Montana indicate that the
wetness indices at the two scales, 2 m and 10 m, are highly correlated to one another
and fairly well correlated to the thickness of mollic horizon. Neither of the wetness
indices or the thickness of mollic horizon correlated well with season average or June
115
soil water variables. Season averages for total water are highly correlated to June
total water data.
The inclusion or exclusion of sites located in the ephemeral stream
r
channel did not have a substantial effect on the results of any data analysis procedure.
Table 42. Correlation (Pearson’s and Spearman Rank) analyses between wetness
indices and thickness of mollic horizon.
Correlation Procedure
Stream Channel In
(n = 67)
Stream Channel Out
(n = 59)
r
P-value
r
P-value
2m
0.4210
0.0000
0.3590
0.0052
10 m
0.6239
0.0000
0.3746
0.0035
2m
0.4857
0.0000
0.4109
, 0.0012
10 m
0.4784
0.0000
0.4023
0.0016
Wetness Index Scale
Pearson’s
Spearman Rank
Based on a knowledge of soil genesis it might be expected that the depth of the
mollic horizon would be correlated to the distribution of water across the landscape
over a long period of time. This may be reflected in the calculation of the wetness
index in that the depth of the mollic horizon in this field is moderately correlated with
both the 2 m and 10 m wetness indices. Wilson et al. (1994) showed that a
combination of image and terrain attributes including the wetness index could predict
the thickness of the mollic horizon for this study area. However, these data indicate
that over three field seasons, including one wet, fallow year (1993), the depth of the
mollic horizon and wetness indices are poorly correlated with measured soil water
content. Moore’s "wetness index" does not appear to be a good indicator of the
116
distribution of soil water and the potential location of zones of saturation in this
catchment. The two basic assumptions incorporated into the calculation of the
wetness index are not met in the soil-landscape environment of the study field. The
assumption that a piezometric head dictates the direction of subsurface flow parallel to
the gradient of subsurface topography indicates either the presence of a water table or
and impeding layer in the soil profile and lateral redistribution of water. Secondly,
the assumption of steady-state conditions is clearly not met in this field possessing
deep, permeable agricultural soils.
The data presented in Table 38 show that the total soil water in 1993 is not
much higher than in the other two years. This perhaps indicates that water is
redistributed to field capacity very quickly in this landscape with deep loess soils.
The distribution of mollic thickness may be more related to near surface processes of
water distribution as a function of time rather than those in the entire soil profile.
117
CHAPTER 4
SUMMARY
The research presented in this thesis is part of larger effort to examine the
possibility of integrating site-specific information about leaching into an on-farm,
field-scale information system. Integrating soil survey, terrain, and microclimate
attributes and image analysis along with information regarding leaching potential into
a GIS format may provide new opportunities for applying leaching potential models to
variable soil landscapes. Subsequent identification of areas susceptible to leaching, or
management zones which consider leaching susceptibility as an important factor,
would provide a basis for more selective N fertilizer application. Site-specific
applications of N fertilizer may improve profitability, and at the same time, reduce
the possibility of ground water contamination.
A 20-ha portion of a farm field in southwestern Montana possessing
considerable topographic relief and substantial variation in both soil and terrain
attributes was selected for this study in an effort to characterize differences in
leaching across the landscape. It was assumed that these soil and topographic
variations would lead to differential infiltration and redistribution of water which, in
turn, would translate into differential leaching of nitrate and bromide. However, the
Springhill site has deep, loess-derived, permeable soils which apparently capture and
118
retain precipitation from rainfall and melting snow without significant lateral
redistribution in the landscape. It appears that the downward component of water
movement overwhelms any tendency towards lateral subsurface flow.
The spatial variability of both the temporal and spatial variability of measured
water content was small. Neutron probe data was by necessity collected around 48
hours after any significant rainfall. Either water is lost quickly (moves rapidly) in
this environment or most soil water content values measured were near field-capacity,
especially below the zone of substantial root uptake. The spatial variability of nitrate
and bromide was also small. The variability of bromide did, however, increase with
time from surface application. This possibly indicates that differences in solute
movement are expressed very slowly in this rainfed field.
Leaching was characterized through three approaches: I) total solute mass and
critical depth at a particular point in time; 2) change in total solute mass and critical
depth with time; and 3) percent of solute unrecoverable with time (based on mass
balance estimates). Nitrate and bromide exhibited statistically different spatial
behavior patterns under field conditions except in a couple time periods. None of the
three approaches yielded stronger or more consistent relationships with the soil, soil
water, or terrain attributes.
Additionally, Moore’s "wetness" index does not appear to be a good indicator
of the distribution of soil water and the potential location of zones of saturation in this
catchment. However, neither of the two basic assumptions incorporated in the
calculation of this index were met at the Springhill site. Based on a knowledge of soil
119
genesis it might be expected that the depth of the mollic horizon would be related to
the distribution of water across the landscape over a long period of time. But, over a
three year period, the thickness of the mollic horizon exhibited only a weak inverse
relationship to measured soil water content.
This study raises fundamental questions regarding spatial and temporal scales
of the interrelationships between environment, nitrogen leaching, and potential
contamination of ground water.
REFERENCES CITED
121
REFERENCES CITED
Aandahl, A.R. 1948. The characterization of slope positions and their influence on
the total nitrogen content of a few virgin soils of western Iowa. Soil Sci. Soc.
Am. Proc. 13:449-454.
Abdalla, N.A., and B. Lear. 1975. Determination of inorganic bromide in soils and
plant tissues with a bromide selective-ion electrode. Commun. in Soil Science
and Plant Analysis. 6:489-494.
Addiscott, T.M. 1977. A simple computer model for leaching in structured soils. J.
Soil Sci. 28:554-563.
Addiscott, T.M. 1982. Simulating diffusion within soil aggregates: A simple model
for cubic and other regularly-shaped aggregates. J. Soil Sci. 33:37-45.
Addiscott, T.M., and R.J. Wagenet. 1985. Concepts of solute leaching in soils: A
review of modelling approaches. J. Soil. Sci. 36:411-424.
Addiscott, T.M., A.P. Whitmore, and D.S. Powlson. 1991. Farming, Fertilizers and
the Nitrate Problem. C.A.B International, Wallingford, UK. 170 pp.
Afyuni' M.M., D.K. Cassel, and W.P. Robarge. 1994. Effect of landscape position
on soil water and com silage yield. Soil Sci. Soc. Am. J. 58:967-974.
Agus, F., and D.K. Cassel. 1992. Field-scale bromide transport as affected by tillage.
Soil Sci. Soc. Am. J. 56:254-260.
Algard, G., J. Jacobsen, and H. Hart. 1994. Montana General Agricultural Chemical
Ground Water Management Plan: Strategies to Protect Ground Water from
Agricultural Chemicals through Education and Information. MDA Doc. 94-
01.
Amoozegar-Fard, A., D.R. Nielsen, and A.W. Wamck. 1982. Soil solute
concentration distributions for spatially-varying pore water velocities and
apparent diffusion coefficients. Soil Sci. Soc. Am. J. 46:3-9.
Anderson, J.L., and J. Bouma. 1977a. Water movement through pedal soils: I.
Saturated flow. Soil Sci. Soc. Am. J. 41:413-418.
122
Anderson, J.L., and J. Bouma. 1977b. Water movement through pedal soils: II.
Unsaturated flow. Soil Sci. Soc. Am. J. 41:419-423.
Anderson, M.G., and T.P. Burt. 1978. The role of topography in controlling throughflow generation. Earth Surf. Proc. 3:331-334.
Bathke, G.R., and D.K. Cassel. 1991. Anisotropic variation of profile.characteristics
and saturated hydraulic conductivity in an Ultisol landscape. Soil Sci. Soc.
Am. J. 55:333-339.
Bathke, G.R., D.K. Cassel, and P.A. McDaniel. 1992. Bromide movement at
selected sites in a dissected Piedmont landscape. J. Environ. Qual. 21:469475.
Bauder, J.W ., B.A. White, and W.P. Inskeep. 1991. Montana extension initiative
focuses on private well quality. J. Soil Water Cons. 20:69-74.
Beckett, P.H.T. 1987. Spatial variability of soil K-status. Proceedings of the 20Ul
Colloqium International Potash Institute, Bern, pp. 357-380.
Bell, J.C., J.A. Thompson, C.A. Butler, and K. McSweeney. 1994. Modeling soil
genesis from a landscape perspective. Transactions: 15th World Congress of
Soil Science, Acapulco, Mexico, July, 1994. Vol. 6A, Commission V:
Symposia.
Bergstrom, L., and R. Johansson. 1991. Leaching of nitrate from monolith lysimeters
of different types of agricultural soils. J. Environ. Qual. 20:801-807.
Beven, K., and M.J. Kirkby. 1979, A physically-based variable contributing area
model of basin hydrology. Hydrol. Sci. Bull. 24:43-69.
Seven, K. 1981. Micro-, meso-, and macroporosity and channeling phenomena in
soils. Soil Sci. Soc. Am. J. 45:1245.
Beven, K., and P. Germann. 1981. Water flow in soil macropores: A combined flow
model. J. Soil Sci. 32:15-29.
Bhatti, A.U., D.J. Mulla, F.E. Koehler, and A H Gurmani. 1991. Identifying and
removing spatial correlation from yield experiments. Soil Sci. Soc. Am. J.
55:1523-1528.
Biggar, J.W ., and D.R. Nielsen. 1976. Spatial variability of the leaching
characteristics of a field soil. Water Resour. Res. 12:78-94.
123
Bouma, J., L.W. Dekker, and C J. Muilwuk. 1981. A field method for measuring
short circuiting in clay soils. J. Hydrology. 52:347-354.
Bouma, J., and P.A. Finke. 1993. Origin and nature of soil resource variability. In:
P.C. Robert, R.H. Rust, and W.E. Larson (eds.), Soil Specific Crop
Management: A Workshop on Research and Development Issues. Amer. Soc.
Agron.; Madison, WL pp. 3-13.
Brady, N.C. 1990. The Nature and Properties of Soils, 10th Edition. MacMillan
Publishing Company, New York, NY. p. 106.
Bremner, J.M ., and C.S. Mulvaney. 1982. Nitrogen-total. In: A.L. Page (ed.),
Methods of Soil Analysis. Part 2. 2nd ed. Agronomy 9:595-624.
Bresler, E. 1967. A model for tracing salt distribution in the soil profile and
estimating the efficient combination of water quality and quantity under
varying field conditions. Soil Sci. 104:227-233.
Bresler, E., H. Bielorai, and A. Laufer. 1979. Field test of solution flow models in a
heterogenous irrigated cropped soil. Water Resour. Res. 15:645-652.
Bresler, E., and G. Dagan. 1979. Solute dispersion in unsaturated heterogenous soil
at field scale. II. Applications. Soil Sci. Soc. Am. J. 43:467-472.
Bresler, E., and G. Dagan. 1981. Convective and pore scale dispersive solute
transport in unsaturated heterogenous fields. Water Resour. Res. 17:16831693,
Bresler, E., and A. Laufer. 1974. Simulation of nitrate movement in soils under
transient unsaturated flow conditions. Israel J. Agric. Res. 23:141-153.
Broadbent, F.E., and F.E. Clark. 1965. Denitrification. In: J.W. Bartholomew, and
F.E. Clark (eds.) Soil Nitrogen. Agronomy Monograph 10. Amer. Soc.
Agron., Madison, WL pp. 347-362.
Brown, N.W., and W.B. Saufferer. 1991, Integration of site specific field data into a
whole farm management system. In: Automated Agriculture for the 21st
Century (Proc. of the 1991 Symposium). ASAE PubL 11-91. pp. 385-393. .
Brubaker, S.C., A.J. Jones, D.T. Lewis, and K. Frank. 1993. Soil properties
associated with landscape position. Soil Sci. Soc. Am. J. 57:235-239.
124
Brubaker, S.C., A J. Jones, K. Frank, and D.T. Lewis. 1994. Regression models for
estimating soil properties by landscape position. Soil Sci. Soc. Am. J.
58:1763-1767.
Bruce, R.R., R.A. Leonard, A.W. Thomas, and W.A. Jackson. 1985. Redistribution
of bromide by rainfall infiltration into a Cecil sandy loam landscape. J.
Environ. Qual. 14:439-445.
Bums, LG. 1972. Leaching studies. Report of the National Vegetable Research
Station for 1971. pp. 39-40.
Bums, LG. 1974. A model for predicting the redistribution of salts applied to a
fallow soil after excess rainfall or evaporation. J. Soil Sci. 25:165-178.
Bums, LG. 1975. An equation to predict the leaching of surface-applied nitrate. J.
Agric. Sci., Cambridge. 85:443-454.
Burrough, P.A. 1993. Soil variability: A late 20lh century view. Soils and Fertilizers.
56:529-562.
Burt, T.P., and D.P. Butcher. 1985. Topographic controls of soil moisture
distributions. J. Soil Sci. 36:469-486.
Butters, G.L. 1987. Field-scale transport of bromide through unsaturated soil. Ph.D.
Dissertation, Univ. of Calif., Riverside. 235 pp.
Caprio, J.M ., D.I. Cooksey, J.S. Jacobsen, G.A. Nielsen, and R.R. Roche. 1994.
MAPS Atlas Version 5.0, Montana State University Extension Bulletin 125,
Bozeman, MT.
Carr, P.M ., G.R. Carlson, J.S. Jacobsen, G.A. Nielsen, and E.O. Skogley. 1991.
Farming soils, not fields: A strategy for increasing fertilizer profitability. J.
Prod. Agric. 4:57-61.
Carsel, R.F., C.N. Smith, L.A. Mulkey, J.D. Dean, and P. Jowise. 1984. User’s
manual for the pesticide root zone model (PRZM): Release I, Rep. EPA600/3-84-109, U. S. Environ. Protect. Agency, Environ. Res. Lab., Athens,
GA. 219 pp.
Clothier, B.E., and D.E. Elrick. 1985. Solute dispersion during axisymmetric threedimensional unsaturated water flow. Soil Sci. Soc. Am. J. 49:552-556.
125
Crowder, B., and C.E. Young. 1988. Managing farm nutrients: Tradeoffs for
surface- and ground-water quality. Econ. Res. Serv., U.S.D.A., Washington,
D.C.
Dagan, G., and E. Bresler. 1979. Solute dispersion in unsaturated heterogenous soil
at field scale. I. Theory. Soil Sci. Soc. Am. J. 43:461-466.
Daniels, R.B., J.W. Gilliam, D.K. Cassel, and L.A. Nelson. 1985. Soil erosion class
and landscape position in the North Carolina Piedmont. Soil Sci. Am. J.
49:991-995.
De Smedt, F., and P.J. Wierenga. 1978. Approximate analytical solution for solute
flow during infiltration and redistribution. Soil Sci. Soc.Am. J. 42:407-412.
Destouni, G. 1993. Stochastic modeling of solute flux in the unsaturated zone at the
field scale. J. Hydrol. 143:46-61.
Dikau, R. 1989. The application of digital relief models to landform analysis in
geomorphology. In: J. Raper (ed.) Three Dimensional Applications in
Geographic Information Systems. Taylor and Francis, New York, NY. pp. 5177.
Er-Raji, M., and L.J. Cihacek. 1994. Erosion effects on selected agronomic
parameters of hard red spring wheat across a landscape. North Dakota Farm
Research Bimonthly Bulletin. 50:26-30.
Foiled, R.F., D.R. Keeney, and R.M. Cruse (eds.). 1991. Managing Nitrogen for
Groundwater Quality and Farm Profitability. Soil Sci. Soc. Am., Madison,
WL
Frissel, M J ., and P. Poelstra. 1967. Chromatographic transport through soils. I.
Theoretical evaluations. Plant Soil. 26:285-301.
Gardner, W.H. 1986. Water Content. In: A. Klute (ed.) Methods of Soil Analysis.
Part I. Agronomy 9:493-544.
Gessler, P.E., I.D. Moore, N.J. McKenzie, and P.J. Ryan. 1993. Soil-landscape
modelling in southeastern Australia. Land Surface Characterization for
Environmental Models Session at Integrating GIS and Environmental
Modelling Conference. Breckenridge, CO.
Hall, G.F., and C.G. Olson. 1991. Predicting variability of soils from landscape
models. In: M J . Mausbach and L.P. Wilding (eds.) Spatial Variability of
Soils and Landforms. SSSA Special Publication No. 28. pp. 9-24.
126
Hanna, A.Y., P.W. Harlan, and D.T. Lewis. 1982. Soil available water as influenced
by landscape position. Agron. J. 74:999-1002.
Hanna, A.Y., P.W. Harlan, and D.T. Lewis. 1983. Effect of slope on water balance
under center-pivot irrigation. Soil Sci. Soc. Am. I. 47:460-764.
Hart, H., and J.S. Jacobsen. 1994. The Montana Agricultural Chemical Ground
Water Protection Act: An Overview. Fertilizer and Pesticide Best Management
Practices. Montana St. Univ. Ext. Service, Bozeman, MT.
He, B., C.L. Peterson, and T. Otawa. 1992. Development of a field scale GIS
database for spatially variable farming management. ASAE Pacific Northwest
Section Meeting Presentation.
Hill, M.J. 1991. Nitrates and Nitrites in Food and Water. Ellis Horwood Limited,
NY. p. 196.
Hutchinson, M.F. 1989. A new procedure for gridding elevation and stream line data
with automatic removal of spurious pits. J. Hydrology. 106:211-232.
Jacobsen, J.S., and G.D. Johnson. 1991. Water quality and agrichemicals in
Montana. Montana St. Univ. Ext. Service Bulletin EB51, Bozeman, MT.
Jardine, P.M ., G.V. Wilson, and R.J. Luxmoore. 1988. Modeling the transport of
inorganic ions through undisturbed soil columns from two contrasting
watersheds. Soil Sci. Soc. Am. J. 52:1252-1259.
Jones, J.A.A. 1986. Some limitations to the a/s index for predicting basin-wide
patterns of soil water drainage. Z. Geomorph, N.F. 60:7-20.
Jones, J.A.A. 1987. The initiation of natural drainage networks. Progress in Phys.
Geog. 11:205-245.
Jury, W.A. 1982. Simulation of solute transport using a transfer function model. 1
Water Resour. Res. 18:363-368.
Jury, W.A. 1985. A review of published studies on spatial variability of soil water
and chemical parameters. EPRI EA-4228.
Jury, W.A., L.A. Stolzy, and P. Shouse. 1982. A field test of the transfer function
model for predicting solute transport. Water Resour. Res. 18:369-375.
127
Jury, W.A., G. Sposito, and R.E. White. 1986. A transfer function model of solute
transport through soil. I. Fundamental concepts. Water Resour. Res. 22:243247.
Jury, W.A., and D.R. Nielsen. 1989. In: R.F. Follett (ed.) Nitrogen Management
and Ground Water Protection. Elsevier Science Publishers B.V., Amsterdam,
Netherlands, pp. 139-157.
Kachanoski, R.G. 1988. Processes in soils-from pedon to landscape. In: T. Roswall et
al. (eds.) Scales and Global Change. John Wiley & Sons, Inc. New York, NY.
Kachanoski, R.G., E. Pringle, and A. Ward. 1992. Field measurement of solute
travel times using time domain reflectomery. Soil Sci. Soc. Am. J. 56:47-52.
Knisel, W.G., R.A. Leonard, and F.M. Davis. 1989. GLEAMS user manual,
Southeast Watershed Research Laboratory, Tifton, GA. 25 pp.
Kung, K.-J.S. 1990. Influence of plant uptake on the performance of bromide tracer.
Soil Sci. Soc. Am. J. 54:975-979.
Larson, W .E., and P.C. Robert. 1991. Farming by soil. In: R. Lai, and T. Pierce
(eds.) Soil management for sustainability. Soil Water Cons. Soc., Ankeny, IA.
pp. 103-112.
Levin, I. 1964. Movement of nitrate through soil columns and undisturbed soil
profiles. Transactions of the 8th International Congress of Soil Science
(Bucharest) IV. pp. 1011-1022.
Lund, R.E. 1993. MSUSTAT Statistical Analysis Package, Version 5.22. Research
and Develbpment Institute, Inc. Montana State. University, Bozeman, MT.
McCool, D.K., and K.G. Renard. 1990. Water Erosion and Water Quality. In: R.P.
Singh, J.F. Parr, and B.A. Stewart (eds.) Adv. in Soil Science: Dryland
Agriculture: Strategies for Sustainability. 13:175-185.
McEachem, K.L., G.A. Nielsen, J.S. Jacobsen, and P.A. McDaniel. 1990.
Application of global positioning system technology for a field-scale
geographic information system. In: Agronomy Abstract. San Antonio, TX.
October 1990. Amer. Soc. Agron., Madison, WL p. 250.
McKenzie, N.J., and D.A. McLeod. 1989. Relationships between soil morphology
and soil properties relevant to irrigated and dryland agriculture. Aust. J. Soil
Res. 27:235-258.
I
I
128
Macy, T.S. 1991. Personal site-specific experiences. Presented to ASAE Automated
Agriculture in the 21st Century Symposium. Dec. 15-17, Chicago, IL.
Mausbach, M.J., D.J. Lytle, and L.D. Spivey. 1993. Application of soil survey
information to soil specific farming. In: P.C. Robert, R.H. Rust, and W.E.
Larson (eds.) Soil Specific Crop Management: A Workshop on Research and
Development Issues. American Society of Agronomy, Madison, WL pp. 5768 .
Miller, M.P., M.J. Singer, and D.R. Nielsen. 1988. Spatial variability of wheat yield
and soil properties on complex hills. Soil Sci. Soc. Am. J. 52:1133-1141.
Montana Climate Center, Montana State University. 1995. Personal communication.
Moore, I.D. 1992. Terrain analysis programs for the environmental sciences: TAPES.
Agric. Systems Info. Tech. 4:37-39.
Moore, I.D., and G.J. Burch. 1986. Modeling erosion and deposition: topographic
effects. Trans. Am. Soc. Agr. Engrs. 29:1624-1630,1640.
Moore, I.D ., G.J. Burch, and D.H MacKenzie. 1988a. Topographic effects on the
distribution of surface soil water and the location of ephemeral gullies.
TRANSACTIONS of the ASAE. 31:1098-1107.
Moore, I.D., P.E. Gessler, G.A. Nielsen, and G.A. Peterson. 1993a. Terrain
analysis for soil-specific crop management. In: P.C. Robert, R H. Rust, and
W.E. Larson (eds.) Soil Specific Crop Management: A Workshop on Research
and Development Issues. American Society of Agronomy, Madison, WL pp.
27-55.
Moore, I.D., P.E. Gessler, G.A. Nielsen, and G.A. Peterson. 1993b. Soil attribute
prediction using terrain analysis. Soil Sci. Soc. Am. J. 57:443-452.
Moore, I.D ., R.B. Grayson, and A.R. Ladson. 1991. Digital terrain modelling: A
review of hydrological, geomorphological, and biological applications. Hydrol.
Processes. 5:3-30.
Moore, I.D., A. Lewis, and J.C. Gallant. 1993c. Terrain attributes: estimation
method and scale effects. In: Jakeman, A.K., M.B. Beck, and J J . McAleer
(eds.) Modelling Change in Environmental Systems. John Wiley & Sons, Inc.
Chichester, pp. 30-38.
Moore, I.D., and J.L. Nieber. 1989. Landscape assessment of soil erosion and
nonpoint source pollution. J. Minnesota Acad. Sci. 55:18-25.
i
129
Moore, I.D., E.M. O’Loughlin, and G J. Burch. 1988b. A contour-based topographic
model for hydrological and ecological applications. Earth Surf. Proc. and
Landforms. 13:305-320.
Moore, I.D., J.C Panuska, R.B. Grayson, and K.P. Srivastava. 1988c. Application of
digital topographic modeling in hydrology. Proc. of the Inti. Symposium on
Modeling Agr., For. and Rangeland Hyd. Dec. 12-13, 1988. Chicago, IL.
pp. 447-461.
Moore, I.D ., A.K. Turner, J.P. Wilson, S.K. Jensen, and L.E. Bank. 1993d. GIS
and land surface-subsurface process modeling. In: M.F. Goodchild, B. Parks,
and L.T. Steyaert (eds.) GIS and Environmental Modeling. Oxford Univ.
Press, New York, NY.
Mulla, D.J. 1990. Assessment of field-scale leaching patterns for management of
nitrogen fertilizer application. In: K. Roth, H. Fuhler, W.A. Jury, and J.C.
Parker (eds.) Field-scale Water and Solute Flux in Soils. Birhauser Verlag
Basel, Germany, pp. 55-63.
Mulla, D.J. 1991. Using geostatistics and GIS to manage spatial patterns in soil
fertility. Automated Agriculture for the 21st Century (Proc. of the 1991
Symposium). ASAE Publ. 11-91. pp. 336-345.
Mulla, D.J. 1993. Mapping and managing spatial patterns in soil fertility and crop
yield. In: Robert, P.C., R.H. Rust, and W.E. Larson (eds.). Soil Specific
Crop Management: A Workshop on Research and Development Issues.
American Society of Agronomy, Madison, WL pp. 15-26.
National Cooperative Soil Survey. 1975. Bridger variant series description.
National Cooperative Soil Survey. 1978. Farland series description.
National Cooperative Soil Survey. 1992. Brodyk tentative series description.
National Cooperative Soil Survey/Soil Conservation Survey. 1994. Soil Survey of
Gallatin County Area, Montana. Unpublished.
National Research Council, Commission of Natural Resources, Panel on Nitrates of
the Coordinating Committee for Scientific and Technical Assessments of
Environmental Pollutants, and Environmental Studies Board. 1978. Nitrates:
An environmental assessment. National Academy of Sciences, Washington,
D.C.
National Research Council. 1989. Alternative Agriculture. National Academy Press,
Washington, D.C. 448 pp.
-
130
Nicholls, P.H., R.H. Bromilow, and T.M. Addiscott. 1982. Measured and simulated
behavior of fluometuron, aldoxycarb and chloride ion in a fallow structured
soil. Pesticide Science 13:475-483.
Nielsen, D.R., and J.W. Biggar. 1962. Miscible displacement. III. Theoretical
considerations. Soil Sci. Soc. Am. Proc. 26:216-221.
Nielsen, D.R., J.W. Biggar, and K.T. Erh. 1973. Spatial variability of fieldmeasured soil-water properties. Hilgardia. 42:215-260.
Nielsen, D.R., 0 . Wendroth, and M.B. Parlange. 1994. Opportunities for examining
on-farm soil variability. Unpublished draft.
Nielsen, G.A. 1994. Personal communication.
Nofziger, D.L., and A.G. Hornsby. 1987. CMLS: Chemical movement through
layered soils model user’s manual. University of Florida, Gainsville, FL.
O’Loughlin, E.M. 1981. Saturation regions in catchments and their relationships to
soil and topographic properties. J. Hydrol. 53:229-246.
O’Loughlin, E.M. 1986. Prediction of surface saturation zones in natural catchments
by topographic analysis. Water Resour. Res. 22:794-804.
Odeh, 1.0.A., D J. Chittleborough, and A.B. McBratney. 1991. Elucidation of soillandform inter-relationships by canonical ordination analysis. Geoderma. 49:132.
Onken, A.B.,, R.S. Hargrove, C.W. Wendt, and O.C. Wilke. 1975. The use of a
specific ion electrode for determination of bromide in soils. Soil Sci. Soc.
Amer. Proc. 39:1223-1225.
Onken, A.B., C.W. Wendt, R.S. Hargrove, and O.C. Wilke. 1977. Relative
movement of bromide and nitrate in soils under three irrigation systems. Soil
Sci. Soc. Am. J. 41:50-53.
Owens, L.B., R.W. Van Keuron, and W.M. Edwards. 1985. Groundwater quality
changes resulting from a surface bromide application to a pasture. J. Environ.
Quality. 14:543-548.
Parker, J.C., and M.T. Van Genuchten. 1984. Determining transport parameters from
laboratory and field tracer experiments. Bulletin No. 84-3. Virginia
Polytechnic Institute and State University, Blacksburg, VA.
131
Pennell, K.D., A.G. Hornsby, R.E. Jessup, and P.S.C. Rao. 1990. Evaluation of five
simulation models for predicting aldicarb and bromide under field conditions.
Water Resoun Res. 26: 2679-2693.
Petersen, C.A. 1991. Precision GPS navigation for improving agricultural
productivity. GPS World. Jan. 1991.
Power, J.F., and J.S. Schepers. 1989. Nitrate contamination of ground water in
North America. Agric. Ecosystems Environ. 26:165-187.
Ratts, P.A.C. 1978. Convective transport of solutes by steady flows. I. General
theory. Agric. Water Management. 1:201-218.
Robert, P.C. 1993. Characterization of soil conditions at the field level for soil
specific management. Geoderma 60:57-72.
Robert, P.C. 1991. Soil specific anhydrous ammonia management system. In:
Automated Agriculture for the 21st Century (Proc. of the 1991 Symposium).
ASAE PubL 11-91. pp. 418-426.
Robert, P C., and R.H. Rust. 1982. Remote sensing for real time agricultural
management. In: C. Johannsen and J. Sanders (eds.) Remote Sensing for
Resource Management. Soil Cons. Soc. Amer., Ankeny, IA. pp. 413-419.
Rose, C.W., F.W. Chichester, J.R. Williams, and J.T. Ritchie. 1982a. A
contribution to simplified models of field solute transport. J. Environ. Qual.
11:146-150.
Rose, C.W., F.W. Chichester, J.R. Williams, and J.T. Ritchie. 1982b. Application
of an approximate analytic method of computing solute profiles with dispersion
in soils. J. Environ. Qual. 11:151-155.
Ruffner, J.A. 1985. Climates of the United States. Vol. I Alabama-New Mexico. 3rd
Ed. Gale Research Company, Detroit, MI.
SAS Institute, Inc. 1988. SAS Procedures Guide and SAS/STAT User’s Guide,
Release 6.03 Edition, Cary, NC. SAS Institute, Inc.
Scotter, D.R. 1978. Preferential solute movement through larger soil voids. I. Some
computations using simple theory. Aust. J. Soil Res. 16:257-267.
132
Schaffer, M.J., A.D. Halverson, and F J . Pierce. 1991. Nitrate leaching and
economic analysis package (NLEAP): Model description and application. In:
R.F. Follett, D.R. Keeney, and R.M Cruse (eds.) Managing Nitrogen for
Groundwater Quality and Farm Profitability. Soil Sci. Soc. Am., Madison,
WL pp. 285-322.
Sharma, M.L., and M. Taniguchi. 1991. Movement of a non-reactive solute tracer
during steady and intermittent leaching. I. Hydrol. 128:323-334.
Silvertooth, J.C., J.E. Watson, J.E. Malcuit, and T.A. Doerge. 1992. Bromide and
nitrate movement in an irrigated cotton production system. Soil Sci. Soc. Am.
J. 56:548-555.
Sims, J. 1994. Personal communication.
Sinai, G., D. Zaslavsky, and P. Golany. 1981. The effect of soil surface curvature on
moisture and yield: Beer Sheba observation. Soil Sci. 123:367-375.
Skidmore, A.K., P.J. Ryan, W. Dawes, D. Short, and E. O’Loughlin. 1991. Use of
an expert system to map forest soils from a geographical information system.
Int. J. Geograph.Inf. Sys. 5:431-445.
Smith, S.J., and R.J. Davis. 1974. Relative movement of bromide and nitrate through
soils. J. Environ. Qual. 3:152-155.
Smith, S J ., L.R. Ahuja, and J.D. Ross. 1984. Leaching of a soluble chemical under
field crop conditions. Soil Sci. Soc. Am. J. 48:252-258.
Spangrud, D.J. 1995. Terrain and soil analysis in support of precision farming.
Unpubl. thesis. Dept, of Earth Science, Montana State University, Bozeman,
MT.
Speight, LG. 1974. A parametric approach to landform regions. Spec. Publ. Inst. Br.
Geogr. 7:213-230.
Starr, LL., and D.E. Glotfelty. 1990. Atrazine and bromide movement through a silt
loam soil. J. Environ. Qual. 19:552-558.
Steenhuis, T.S., S. Pacenka, and K.S. Porter. 1987. MOUSE: A management model
for evaluating groundwater contamination from diffuse surface sources aided
by computer graphics. Appl. Agr. Res. 2:277-289.
133
Stone, LR., J.W. Gilliam, D.K. Cassel, R.B. Daniels, L.A. Nelson, and H J. Kleis.
1985. Effect of erosion and landscape position on the productivity of Piedmont
soils. Soil Sci. Soc. Am. J. 49:987-991.
Tanji, K.K., L.D. Doneen, G.V. Ferry, and R.S. Ayers. 1972. Computer simulation
analysis on reclamation of salt-affected soils in San Joaquin Valley, California.
Soil Sci. Soc. Am. Proc.' 36:127-133.
Terkeltroub, R.W., and K.L. Babcock. 1971. A simple method for predicting salt
movement through soil. Soil Sci. 111:182-187.
Thomas, G.W. 1970. Soil and climatic factors which affect nutrient mobility. In:
O.P. Engelstad (ed.) Nutrient Mobility in Soils: Accumulation and Losses.
Soil Sci. Soc. Am., Madison, WL
Thomas, G.W., R.B. Phillips, and V.L. Quisenberry. 1978. Characterization of water
displacement in soils using simple chromatography theory. J. Soil Sci. 29:3237.
■
Tillotson, W.R., and R.J. Wagenet. 1982. Simulation of fertilizer nitrogen under
cropped situations. Soil Sci. 133:133-143.
Troeh, F.R. 1964. Landform parameters correlated to soil drainage. Soil Sci. Soc.
Proc. 28:808-812.
Trudgill, S.T., A. Pickles, and K.R.J. Smettem. 1983. Soil water residence time and
solute uptake. 2. Dye tracing and preferential flow predictions. J. Hydrology.
62:279-285.
Van de Pol, R.M., P J . Wierenga, and D.R. Nielsen. 1977. Solute movement in a
field soil. Soil Sci. Soc. Am. J. 41:10-13.
Van Genuchten, M.Th., and P J . Wierenga. 1976. Mass transfer studies in porous
/ sorting media. I. Analytical solutions. Soil Sci. Soc. Am. J. 40:473-480.
Veseth, R., and C. Montagne. 1980. Geologic parent material of Montana soils.
Montana Agric. Exp. Sta. Bulletin 721, Montana St. Univ., Bozemen, MT and
USDA-SCS, Bozeman, MT.
Wagenet, R J ., and J.L: Hutson. 1987. LEACHM: A finite-difference model for
simulating water, salt, and pesticide movement in the plant root zone,
continuum 2. New York State Resources Institute. Cornell University, Ithaca,
NY.
134
Wagenet, R.J., and B.K. Rao. 1983. Description of nitrogen movement in the
presence of spatially variable soil hydraulic properties. Agric. Water
Management. 6:227-242.
Walker, P.H., G.F. Hall, and R. Protz. 1968. Relation between landform parameters
and soil properties. Soil Sci. Soc. Am. I. 32:101-104.
Webster, R. 1977. Canonical correlation in pedology: how useful? J. Soil Sci.
28:196-221.
White, R.E. 1985. The influence of macropores on the transport of dissolved and
suspended matter through soil. Adv. Soil Sci. 3:95-120.
White, R.E., J.S. Dyson, R.A. Haigh, W.A. Jury, and G. Sposito. 1986. A transfer
function model of solute transport through soil. 2. Illustrative application.
Water Resour. Res. 22:248-254.
Wilson, J.P., D.J. Spangrud, M.A. Landon, J.S. Jacobsen, and G.A. Nielsen. 1994.
Mapping soil attributes for site-specific management of a Montana farm field.
In: G.E. Meyer and J.A. DeShazer (eds.), Proceedings of Optics in
Agriculture, Forestry, and Biological Processing. 2345:324-335.
Winteringham, F.P.W. 1984. Soil and Fertilizer Nitrogen. International Atomic
Energy Agency. Technical Reports Series No. 244. Vienna, Austria, p.106.
Wraith, J.M. 1994. Personal communication.
Wraith, J.M ., and W. Inskeep. 1994. Personal communication.
Wylie, B.K., M.J. Shaffer, M.K. Brodahl, D. Dubois, and D.G. Wagner. 1994.
Predicting spatial distributions of nitrate leaching in northeastern Colorado. J.
Soil Water Cons. 49:288-293.
Zaslavsky, D., and A.S. Rogowski. 1969. Anisotropy and infiltration in soil profile
development. Soil Sci. Soc. Am. Proc. 33:594-599.
Zaslavsky, D., and G. Sinai. 1981a. Surface hydrology: !-Explanation of phenomena.
J. Hydraulics Div. ASCE 107:1-16.
Zaslavsky, D., and G. Sinai. 1981b. Surface hydrology: Ill-Causes of lateral flow. J.
Hydraulics Div. ASCE 107:37-52.
Zaslavsky, D., and G. Sinai. 1981c. Surface hydrology: IV-Flow in sloping, layered
soil. J. Hydraulics Div. ASCE 107:53-64.
135
Zaslavsky, D., and G. Sinai. 198Id. Surface hydrology: V-In-surface transient flow.
J. Hydraulics Div. ASCE 107:65-93.
I
APPENDICES
APPENDIX A
PROCEDURES FOR PLANT TISSUE AND SOIL BROMIDE ANALYSIS
138
Soil Bromide Extraction Technique
1. Make eluant solution consisting of 2 liter mixed solution of 0.1908 g/L NaCO3 and
0 . 1428.g/L NaHCO3
2. Add 20 ml eluant solution to 10 g prepared soil (air-dried and ground)
3. Shake soil/solution mixture for one hour
4. Filter soil/solution mixture through Whatman glass filter
5. Filter solution from #4 through Supor-450 0.45um membrane filter
6. Analyze filtered solution on ion chromatograph
Plant Tissue Analysis for Bromide CBr-I
1. Make standard solutions
2. Create standard curve
3. Calibrate bromide electrode prior to each run
4. Add 25 ml DDH2O with ISA to I gm dry, plant tissue
5. Shake for 30 minutes (setting #7)
6. Homogenize for 3 minutes (setting 3.5)
7. Centrifuge at 10,000 rpm for 6 minutes
8. Filter through double thickness #1 Whatman paper
9. Calibrate bromide electrode and take millivolt (mv) readings (stir solution with
magnetic stirrer)
APPENDIX B
LEACHIND .BAS OUTLINE AND PROGRAM
140
LEACHIND.BAS
Leaching Index Generation Program
Written by Dr. J.M. Wraith and M.A. Landon
DECLARE subroutine and functions
DIMENSION arrays
DEFINE in and out file names
OPEN in and out files
READ flag jd , depth (inches), and concentration (mg/kg) values for each flagjd
(assign a value of 999 for missing (non-measured) depth and concentation values)
from data files
CONVERT depth values from inches to centimeters
FIT Polynomial Function (N-l degree) to measured data at each flag jd , and
INTERPOLATE data at 5-cm depth increments from the fitted equation at each
flag jd
SET all interpolated data points between last two measured points to zero, for cases
where last two measured points were equal to zero
NUMERICAL INTEGRATION of the N-I polynomial-fitted (i.e. 5-cm increment)
data at each flag jd using Newton’s Forward Integration (2nd order) Function,
Bessel’s Integration (3nd order) Function, and Newton’s Backward Integration (2nd
order) Function for equally-spaced depth values
CALCULATE total solute load (TSolMass, mg/kg) and depth at which 80% of the
total load is above (CritDepth, cm) at each flagjd
PRINT resultant data to out files*
*NOTE: Need to convert data to volume basis rather than mass basis in order to
validly integrate and plot concentration vs. depth. Not a problem if just want to
calculate CritDepth.
141
Figure 14. LEACHIND.BAS program.
***********************************************************************
LEACHIND.BAS
July 27, 1994
Leaching Index Generation Program
Written by Dr. Jon Wraith and Melissa Landon
Depth array contains the measurement depths (cm)
Cone array contains the measured solute concentration profile
NumPts array contains the number of points per site location
MaxDepth array contains the maximum depth measured at each site location
TSolMass array contains the total solute mass
CritDepth array contains the depth at which 80% of solute is below
DepthCon array contains the fitted solute concentration profile
ID is the site location identification number
Depth is the soil depth at the midpoint of the soil core increment
Cone is the solute concentration for the soil depth increment
N is a number of points counter i***********************************************************************
DECLARE SUB PolInt (XA!(), YA!Q, N!, X!, Y!, DY!)
DECLARE FUNCTION BESSINT3! (DeltX!, FMINl!, FO!, F l!, F2!)
DECLARE FUNCTION NBAKINT2! (DeltX!, FMIN2!, FM INl!, PO!)
DECLARE FUNCTION NF0RINT2! (DeltX!, FO!, Fl!, F2!)
DIM Depth(70, 7), Conc(70, 7), PredConc(70, 40)
DIM NumPts(70), MaxDepth(70), PredConcError(70, 40)
DIM TSolMass(70), CritDepth(70), DepthCon(70, 36)
CLS
Tnfil$ = "c:\melissa\brdats93"
INPUT "Enter file name (no extension)"; Infil$
InfileS =
OutfilelS
Outfile2$
OutfileSS
OPEN
OPEN
OPEN
OPEN
InfilS +
= InfilS
= InfilS
= InfilS
".csv"
+ ".tst"
+ ".sum"
+ ".fit"
InfileS FOR INPUT AS #1
OutfilelS FOR OUTPUT AS #2
Outfile2$ FOR OUTPUT AS #3
Outfile3$ FOR OUTPUT AS #4
142
Figure 14 continued. LEACHIND.BAS program.
FOR ID = I TO 70
N = O
FOR Depth % = I TO 7
INPUT #1, ID
INPUT #1, D$
INPUT #1, c$
IF D$ = "" THEN
Depth (ID, Depth %) = 999
Conc(ID, Depth %) = 999
ELSE
Depth(ID, Depth %) = VAL(D$)
Conc(ID, Depth %) = VAL(c$)
END IF
IF Depth(ID, Depth %) = 999 THEN
NumPts(ID) = N
IF Depth % = 7 THEN EXIT FOR
FOR D = Depth % + I TO 7
INPUT #1, ID
’read data pts., & just assign 999’s
INPUT #1, D$
INPUT #1, c$
Depth(ID) D) = 999
Conc(ID) D) = 999
NEXT D
EXIT FOR
’go to next tube ID
ELSE
N = N + I
IF N = I THEN
TotalDepth = Depth(ID, Depth %)
Depth(ID) Depth%) = Depth(ID, Depth%) / 2
ELSEIF N > I THEN
IastDepth = TotalDepth
TotalDepth = Depth(ID, Depth %)
Depth(ID, Depth%) = IastDepth + (TotalDepth - IastDepth) / 2
END IF
END IF
NEXT Depth %
NumPts(ID) = N
NEXT ID
143
Figure 14 continued. LEACHIND.BAS program.
’convert depth midpoints to cm
FOR ID = I TO 70
FOR Depth % = I TO 7
IF Depth(ID, Depth%) = 999 THEN
MaxDepth(ID) = Depth(ID, Depth % - I)
EXIT FOR
ELSE
Depth(ID, Depth%) = Depth(ID, Depth%) * 2.54
END IF
MaxDepth(ID) = Depth(ID, Depth%)
NEXT Depth %
NEXT ID
’convert to cm
’test file input and manipulation routines
FOR ID = I TO 70
FOR Depth % = I TO 7
PRINT #2, USING "## ###.# ###.#### ###.##"; ID; Depth(ID,
Depth%); Conc(ID, Depth%); MaxDepth(ID)
NEXT Depth %
NEXT ID
CLOSE #2
’END
************************************************************************
’Fit function to measured data, and interpolate data at 5-cm
’depth increments from the fitted equation.
FOR ID = I TO 70
’tubes
Nl% = NumPts(ID)
IF NI % = 0 THEN GOTO 222 ’skip over PolInt sub (N will be zero)
FOR D = I TO NumPts(ID) ’load depth and cone, values into temporary arrays '
Dpth(D) = Depth (ID, D)
Cnc(D) = Conc(ID, D)
NEXT D
’
CALL Spline(Dpth(), CncQ, Nl %, 1E+31, 1E+31, DerivSplinQ)
FOR D = O T O INT(MaxDepth(ID) / 5) ’no. depths in 5 cm increments
PredConc(ID, D) = 0
’initialize to zero
Dpthl = D * 5
’ depth in cm
144
Figure 14 continued. LEACHIND.BAS program.
’
222
CALL Splint(Dpth(), CncQ, DerivSplinO, Nl %, D pthl, PC)
CALL PolInt(DpthO, CncO, NumPts(ID), DpthI, PC, PCerror)
PredConc(ID, D) = PC
PredConcError(ID, D) = PCerror
NEXT D
’if no data for this tube, don’t go to PolInt sub
NEXT ID
I***********************************************************************
’Set all interpolated data points between last 2 measured pts. = zero, for
’cases where last 2 measured points = 0.
FOR ID = I TO 70
IF NumPts(ID) = 0 THEN GOTO 333
IF (Conc(ID, NumPts(ID) - I ) = 0 AND Conc(ID, NumPts(ID)) = 0) THEN
D l = Depth(ID, NumPts(ID) - I) ’depth in cm of second-from-bottom
D2 = INT(D1 /5 )
D3 = D l / 5
IF D2 < D3 THEN D2 = D2 + I
FOR D4 = D2 TO INT(MaxDepth(ID) / 5)
PredConc(ID, D4) = 0
’set all to zero
NEXT D4
END IF
333
’if no data for a tube, go to next tube
NEXT
i***********************************************************************
’Numerical integration of the spline-fitted (i.e. 5-cm depth increment) data
’to calculate total solute load and depth at which 80% of total load is above.
’NOTE: need to convert data to volume basis rather than mass basis in order to
’ validly integrate, if want to plot Concentration vs. Depth.
’ Not a problem if just want Critical Depth.
145
Figure 14 continued. LEACHIND.BAS program.
DeltX = 5
’depth increment in cm
FOR ID = I TO 70
TSolMass(ID) = 0
’initialize
FOR D = I TO INT(MaxDepth(ID) / 5) + I ’skip depth zero (sometimes
extreme)
IF D = I THEN
Cnc = NFORINT2(DeltX, PredConc(ID, D), PredConc(ID, D + I),
PredConc(ID, D + 2))
ELSEIF D < 35 THEN ’Bessel’s requires 2 points "ahead o f
Cnc = BESSINT3(DeltX, PredConc(ID, D - I), PredConc(ID, D),
PredConc(ID, D + I), PredConc(ID, D + 2))
ELSE
Cnc = NBAKINT2(DeltX, PredConc(ID, D - 2), PredConc(ID, D - I),
PredConc(ID, D))
END IF
DepthCon(ID, D) = Cnc
TSolMass(ID) = TSolMass(ID) + Cnc
NEXT D
’Determine depth at which 80% of solute load is above
SumConc = O
FOR D = I TO INT(MaxDepth(ID) / 5) + I
IF TSolMass(ID) = O THEN
CritDepth(ID) = O
EXIT FOR
END IF
SumConc = SumConc + DepthCon(ID, D)
IF SumConc > = .8 * TSolMass(ID) THEN
CritDepth(ID) = D * 5
EXIT FOR
END IF
NEXT D
NEXT ID
146
Figure 14 continued. LEACHIND.BAS program.
>***********************************************************************
’Print out Resultant Data to Outfiles
PRINT #3, "ID TSolMass CritDepth"
PRINT #4, "ID Depth Cone"
FOR ID = I TO 70
PRINT #3, USING "## MffM.# ###.#"; ID; TSolMass(ID); CritDepth(ID)
FOR Depth % = I TO 36
D = Depth % * 5
PRINT #4, USING "## Mft.1t Mft.ft" \ ID; D; DepthCon(ID, Depth%)
NEXT Depth %
NEXT ID
CLOSE
END
************************************************************************
’BESSEL’S INTEGRATION (3rd Order) FUNCTION:
’ Calculates Finite Difference Approx, to Integral for 3rd Order
’ Approximation (Order Polynomial) based on Bessel’s formula.
’
Assumes: K=O, Equally-spaced X-Values.
’ ARGUMENTS PASSED IN are:
’
DELTX = delta-x increment (may use I if delta-x = delta-y)
’
FM IN l5FO,FI,F2 are Y=F(X) points to use for 3rd order
’
approx.
’ RETURNS the value of the integral from FO to FI.
9
FUNCTION BESSINT3 (DeltX, FM INl, FO, F I, F2)
BESSINT3 = DeltX * (13 * FO + 13 * F l - F2 - FMINI) / 24
END FUNCTION
147
Figure 14 continued. LEACHIND.BAS program.
************************************************************************
’NEWTON’S BACKWARD INTEGRATION (2nd order) FUNCTION:
’ Calculates Finite Difference Approx, to Integral for 2nd Order
’ Approximation (Order Polynomial) based on Newton’s Backward formula.
’
Assumes: K=O, Equally-spaced X-Values.
’ ARGUMENTS PASSED IN are:
’
DELTX = delta-x increment (may use I if delta-x = delta-y)
’
FMIN2,FM INljFO are Y=F(X) points to use for 2nd order
’
approx.
’ RETURNS the value of the integral from FMINl to F0.
FUNCTION NBAKINT2 (DeltX, FMIN2, FM IN l, PO)
NBAKINT2 = DeltX * (8 * FMINl + 5 * FO - FMIN2) / 12
END FUNCTION
************************************************************************
’NEWTON’S FORWARD INTEGRATION (2nd Order) FUNCTION:
’ Calculates Finite Difference Approx, to Integral for 2nd Order
’ Approximation (Order Polynomial) based on Newton’s Forward formula.
’
Assumes: K =0, Equally-spaced X-Values.
’ ARGUMENTS PASSED IN are:
’
DELTX = delta-x increment (may use I if delta-x = delta-y)
’
F0, F I, F2 are Y=F(X) points to use for 2nd order
’
approx.
’ RETURNS the value of the integral from FO TO FI.
FUNCTION NFORINT2 (DeltX, F0, F I, F2)
NFORINT2 = DeltX * (8 * F l + 5 * FO - F2) / 12
END FUNCTION
SUB PolInt (XA0, YA(), N, X, Y, DY)
’Given arrays XA and YA, each of length N, and given a value X, this
’routine returns a value Y, and an error estimate DY. IF P(X) is the
’polynomial of degree N - I such that P(XAi) = YAi, i = 1,...,N , then
’the returned value Y = P(X).
148
Figure 14 continued. LEACHIND.BAS program.
DIM c(N), D(N)
NS = I
DIF = ABS(X - XA(I))
’Find the index NS of closest table entry.
FOR I = I TO N
DIPT = ABS (X - XA(I))
IF DIPT < DIF THEN
NS = I
DIF = DIPT
END IF
’Initialize the tableau of C s and D’s.
C(I) = YA(I)
D(I) = YA(I)
NEXT I
’This is the initial approximation to Y.
Y = YA(NS)
NS = NS - I
’For each column of the tableau,
FOR M = I TO N - I
’loop
over the current C’s and D ’s and update.
FOR I = I TO N - M
HO = XA(I) - X
HP = XA (I + M) - X
W = c(I + I) - D(I)
DEN = HO - HP
IF DEN = 0! THEN PRINT "Abnormal exit": EXIT SUB
DEN = W / DEN
’This error can occur only if two input XA’s
D(I) = HP * DEN
’are (to within roundoff) identical.
c(I) = HO * DEN
’The C s and D’s are updated.
NEXT I
IF 2 * NS < N - M THEN ’After each column in the tableau is completed,
DY = c(NS + I)
’decide the correction, C or D, to add to
’accumulating
value of Y, ie. which path to take
ELSE
’through
the
tableau, forking up or down. Done
DY = D(NS)
’in such a way as to take the most "straight"
NS = NS - I
’line
route through the tableau to its apex,
END IF
’updating NS accordingly to keep track of
Y = Y + DY
’location. This route keeps the partial
NEXT M
’approximations centered on the target X. The
ERASE D, c
. ’last DY added is thus the error indication.
END SUB
I
149
APPENDIX C
INITIAL VALUES FOR SOIL AND TERRAIN ATTRIBUTES
150
Table 43. Initial soil test values.
Flag
ID
pH
(1:2)
I
2
5.8
6.0
5.7
3
4
5
6
7
EC
(1:2)
(dS nr1)
0.06
0.06
0.05
0.07
0.08
0.05
0.06
0.06
OM
(%)
K
(mg kg"1)
3.52
2.33
310
222
2.90
3.14
3.44
2.85
4.00
3.66
Olsen
P
(mg kg"1)
15
15
57
62
268
19
266
388
238
386
28.7
44.7
30.2
64.8
326
324
302
51.6
48.8
52.2
49
63
18
20
33
76
64
85
51
131
85
9
10
5.8
0.06
3.08
3.47
11
7.1
0.23
2.09
222
24.7
12
13
14
15
6.0
5.9
6.1
6.0
0.05
0.04
0.05
0.05
2.65
3.21
3.55
3.24
274
260
286
294
16
17
6.0
5.9
0.06
0.07
18
5.9
0.06
19
6.5
0.09
2.31
3.79
3.93
2.77
20
21
22
23
24
8.2
6.4
5.8
6.5
0.12
0.09
0.08
0.05
1.69
2.43
4.84
2.00
0.06
25
5.8
6.0
0.06
3.58
4.32
26
6.0
0.05
27
5.9
0.04
0.06
0.05
6.0
28
* Value not available.
Depth
to
CaCO3
(cm)
37.9
27.2
*
6.3
6.2
6.0
5.9
6.1
5.8
8
Thickness
of Mollic
Horizon
(cm)
30
43
>180
107
>180
51.0
35.9
37.0
36.8
66
27
37
53
34
210
392
27.0
62.1
16
22
352
51.0
25
48
77
67
238
174
302
490
184
392
25.4
80
99
23.3
55.0
63.8
27.7
27
>180
107
46
>180
466
52.9
55.2
15
55
52
16
76
114
3.37
314
27.2
53
>180
3.26
3.21
278
242
32.0
41
87
33.8
15
48
79
86
88
67
>180
151
Table 43 continued. Initial soil test values.
Flag
ID
pH
(1:2)
EC
(1:2)
(dS n r1)
29
30
5.9
6.0
7.1
0.09
0.07
0.04
4.92
4.83
1.08
6.0
0.06
4.18
39
6.5
0.11
0.07
0.09
0.09
0.13
0.06
0.08
2.57
38
7.6
6.4
6.3
6.2
7.8
5.8
40
6.3
0.04
2.82
41
42
43
44
45
46
47
6.1
6.4
6.1
6.4
6.2
0.05
0.07
0.08
0.06
0.05
0.06
48
49
31
32
33
34
35
36
37
Thickness
of Mollic
Horizon
(cm)
Depth
to
CaCO3
(cm)
8.1
62
80
24
>180
>180
>180
59.0
48
>180
69.3
35.6
32.1
32.4
64.7
60
53
28
35
29
42
18
>180
>180
112
67
46
>180
69
19
50
Olsen
P
(mg kg"1)
OM
(%)
K
(mg kg'1)
662
564
166
522
71.6
57.2
320
410
270
324
230 .
388
248
3.93
3.55
4.04
3.41
4.29
4.47
48.9
34.6
26.1
4.05
2.68
4.25
3.41
3.43
4.27
236
300
326
450
224
172
340
0.10
5.55
844
92.1
5.8
0.08
5.55
388
6.2
6.6
6.2
0.07
5.6
0.08
3.96
4.38
4.45
4.95
55
5.9
6.4
0.09
0.07
2.61
4.35
56
6.2
0.05
57
6.1
58
6.0
50
51
53
54
6.3
6.3
26
50
109
26
20
60
- 117
HO
126
>180
56
79
112
58.3
50
>180
304
37.1
20
84
286
344
48.3
52.8
81.5
23
69
27.7
58
34
94
149
>180
74
- 502
93.1
126
>180
226
32.2
52
>180
0.05
2.91
3.95
294
63.1
54
0.12
3.98
218
35.8
36
118
64
0.11
0.06
466
306
33.0
38.3
69.3
28.3
17.8
37.1
.
>180
152
Table 43 continued. Initial soil test values.
EC
(1:2)
(dS nr1)
OM
(%)
K
(mg kg'1)
0.11
0.04
0.04
0.06
3.09
2.78
1.80
3.89
292
272
170
318
0.06
3.09
250
8.1
0.16
216
8.3
6.3
6.2
5.8
0.11
0.05
0.09
0.05
2.20
2.28
3.29
4.18
4.00
Flag
ID
pH
(1:2)
59
60
61
62
63
8.2
6.2
6.6
5.9
6.0
66
67
68
69
70
188
208
454
182
Olsen
P
(mg k g 1)
Thickness
of Mollic
Horizon
(cm)
Depth
to
CaCO3
(cm)
70.1
52.2
15.8
57.2
32.2
118
40
14
51
48
>180
>180
51
95
25.9
22.0
27.4
64.2
63.3
10
83
10
31
45
64
43
70
59
>180
>180
T a b le 4 4 . M e a s u r e d and c a lc u la te d terrain a ttrib u tes.
Profile
Curvature
(m nr1)
1529.03
11.340
254.922
63.2
. 6.323
1.893
-0.191
5076878.0
5076884.0
5076820.0
5076821.0
5076823.0
5076830.0
5076764.0
5076765.0
5076765.0
5076771.0
5076707.0
5076709.0
5076710.0
5076712.0
1531.08
1527.35
1527.11
1528.81
1529.62
1522.20
1525.02
10.378
18.160
3.578
2.549
5.447
13.552
4.622
7.406
8.146
20.865
. 11.710
6.757
129.134
117.553
303.024
244.440
133.512
117.227
211.283
177.678
159.897
22.0
109.0
13.9
14.6
12.7
59.6
35.2
5.357
6.397
5.962
6.350
5.452
6.086
1.786
0.216
-17.896
-17.695
-18.245
-4.776
-7.771
-4.740
-0.025
-0.323
-0.030
0.071
-0.524
0.406
-0.310
0.051
5076713.0
5076650.0
1518.67
1513.10
Y Coordinate
(UTM m)
I
498364.8
5076876.0
2
3
4
5
6
7
8
498420.3
498458.8
498311.3
498365.6
498422.4
498468.5
498319.8
498372.7
498425.4
49.8479.7
498274.0
• 498321.8
498375.5
498428.0
498482.1
498277.2
16
17
Plan
Curvature
(m m"1)
Aspect
(deg)
X Coordinate
(UTM m)
9
10
11
12
13
14
15
Wetness
Index
Slope
1%)
Flag
ID
Elevation
(m)
Specific
Catchment
Area
(m2 m'1)
1525.50
1525.18
1518.77
1519.17
1519.32
1519.29
1519.12
.
13.425
10.842
12.864
8.416
64.295
187.853
177.455
162.442
194.421
138.940
194.449
50.5
54.1
55.6
73.5
128.2
73.8 .
86.4
21.4
149.7
6.635
6.525
6.498
5.585
6.442
7.548
6.309
6.681
5.114
7.484
1.404
3.103
3.113
5.343
-2.749 .
2.220
-9.174
0.296
0.273
0.753
0.446
0.982
-0.332
0.255
-0.302
0.045
T a b le 4 4 c o n tin u e d . M e a su r e d and c a lc u la te d terrain a ttrib u tes.
Elevation
(m)
Plan
Curvature
(m m"1)
Profile
Curvature
(m m'1)
Aspect
(deg)
1499.91
7.760
16.494
14.378
23.359
4.314
6.642
11.312
9.774
5.450
17.913
15.523
5.615
3.501
4.117
8.840
177.045
147.755
153.791
166.886225.939
154.592
163.568
187.052
213.398
175.197
154.838
193.913
178.363
204.386
135.917
91.7
56.3
90.7
86.6
175.4
54.9
997.5
143.7
174.1
112.5
50.5
1978.1
387.4
167.3
236.5
7.075
5.833
6.447
5.915
8.310
6.717
9.085
7.293
8.069
6.443
5.785
10.470
9.311
8.310
7.892
-0.963
-3.385
4.940
-0.289
1.490
3.061
21.572
3.710
-0.831
4.503
-2.135
3.395
2.070
-1.097
5.741
-0.015
0.104
1.086
0.644
0.155
-0.113
1.187
0.589
-0.084
-0.200
0.079
0.188
-0.032
-0.284
0.427
1499.55
1498.05
11.577
7.338
109.687
192.995
87.9
46.3
6.632
6.447
-6.451
-1.521
-0.268
-0.187
Y Coordinate
(UTM m)
31
32
498323.4
498377.7
498429.5
498486.7
498275.3
498325.9
498380.9
498434.3
498488.6
498278.3
498327.8
498383.0
498435.4
498487.7
498276.9
5076653.0
5076654.0
5076656.0
5076671.0
5076587.0
5076592.0
5076597.0
5076601.0
5076606.0
5076535.0
5076538.0
5076542.0
5076546.0
5076550.0
5076477.0
1513.99
1513.68
1512.59
1510.89
1509.61
1510.07
1503.58
1503.82
1505.52
1505.86
1504.79
1500.12
1501.20
1502.34
33
34
498332.4
498384.0
5076485.0
5076488.0
18
19
20
21
22
23
24
25
26
27
28
29
30
Wetness
Index
Slope
(%)
X Coordinate
(UTM m)
Flag
ID
Specific
Catchment
Area
(m2 m"1)
T a b le 4 4 c o n tin u e d . M e a su r e d an d c a lc u la te d terrain a ttr ib u te s.
Flag
ID
X Coordinate
(UTM m)
35
36
37
38
39
40
41
42
43
44
45
46
47
48
498439.1
498487.8
498276.5
498332.2
498385.8
498440.5
498487.8
498275.8
498332.9
498386.3
498440.5
498488.7
; 498275.3
498330.4
49
50
498386.4
51
498442.3
498489.3
Y Coordinate
(UTM in)
5076491.0
5076493.0
5076428.0
5076431.0
5076434.0
5076436.0
5076438.0
5076367.0
5076374.0
5076377.0.
5076378.0
5076380.0
5076316.0
5076318,0
5076321.0
5076324.0 '
5076326.0
Elevation
(m)
1501.82
1505.92
1497.02
1496.48
1496.95
1504.34
1505.57
1491.86
1493.11
1497.96
1500.69
1499.68
1489.85
1493.40
1497.29
1497.44
1495.69
Slope
(%)
18.674
10.333
7.510
16.015
15.076
6.323
4.851
13.844
6.668
9.237
7.301
10.159
1.503
10.533
3.970
5.582
10.958
Aspect
(deg)
Specific
Catchment
Area
(m2 m"‘)
Wetness
Index
313.481
332.319
233.664
102.257
286.572
247.694
171.703
184.557
169.196
258.447
179.215
155.579
273.814
295.592
196.837
48.6
22.5
337.6
35.5
51.2
18.7
13.4
31.4
7653.2
43.4
53.8
105.6
8717.5
58.6
22.0
5.562
5.383
8.411
5.401
5.828
5.689
5.621
5.424
11.651
6.152
6.602
6.946
13.271
6.321
149.300
157.179
74.7
132.8
7.199
7.100
6.317
Plan
Curvature
(m in 1)
2.759
0.526
7.622
1.517
-0.990
-6.720 '
-15.593
-3.292
7.300
0.267
4.121
0.139
25.310
3.658
-5.900
-1.121
1.270
Profile
Curvature
(m m"1)
0.422
-0.437
0.067
0.180
0.231
-0.005
-0.900
0.968
0.649
-0.084
0.416
0.154
0.040
-0.192
0.004
-0.027
0.207
T a b le 4 4 c o n tin u e d . M e a s u r e d and c a lc u la te d terrain a ttrib u tes.
Flag
ID
53
54
55
56
57
58
59
60
61
62
63
66
67
68
69
70
X Coordinate
(UTM m) .
498315.3
498401.4
498302.4
498453.1
498399.6
498459.9
498411.4
498348.9
498344.3
498299.8
498342.2
498296.8
498297.9
498397.1
498371.3
498368.2
Y Coordinate
(UTM m)
5076872.0
.5076894.0
5076343.0
5076799.0
5076784.0
5076290.0
5076634.0
5076585.0
5076551.0
5076678.0
5076688.0
5076431.0
5076400.0
5076404.0
5076459.0
5076409.0 '
Elevation
(m)"
Slope
(%)
Aspect
(deg)
1523.68
1532.08
1491.16
1524.93
1526.70
1495.31
1509.99
1505.79
1504.95
1516.34
1517.59
1499.96
1497.12
3.762
2.616
3.835
20.976
7.748
4.893
12.806
20.373
18.112
9.949
11.575
11.611
12.572
1498.89
1495.53
1496.37
10.078
5.092
265.426
225.000
247.782
87.131
198.435
158.416
193.548
119.396
143.984
188.089
171.555
257.819
197.354
283.485
172.666
15.099
290.956
Specific
Catchment
Area
(m2 in 1)
895.5
22.7
8189.6
33.6
52.3
44.1
132.1
106.7
30.1
112.3
103.1
10.8
17.2
59.1
6417.3
68.2
Wetness
Index
10.078
6.766
12.272
5.076
6.515
6.804
6.939
6.261
5.1.13
7.029
6.792
4.533
4.919
6.374
11.744
6.113
Plan
Curvature
(m m"1)
27.339
-3.344
3.382
0.780
1.846
-3.887
-0.334
1.882
-3.847
-0.155
-0.065
-7.047
-7.638
1.773
28.838
0.473
Profile
Curvature
(m m"1)
0.201
-0.172
0.050
-0.772
0.651
0.040
0.012
0.345
-0.079
0.084
-0.061
-1.844
-0.009
-0.245
0.908
0.086
157
APPENDIX D
MULTIPLE REGRESSION RESULTS
158
Table 45. Multiple regression results for nitrogen critical depth in Spring 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out (n = 57)
Intercept
TW2592
TW15092
160.0
-20.3
1.9
Intercept
TW15092
TW2592
110.7
1.9
-13.2
0.09
0.09
0.20
0.11
Stream Channel In (n = 65)
0.13
0.07
0.13
0.20
5.32
7.86
0.0249
0.0070
9.37
0.0032
0.0185
5.85
Table 46. Multiple regression results for change in nitrogen solute mass
from Fall 1992-Spring 1993.
Variable
Parameter
Estimate
Partial r2
Model R2.
Prob > F
F
Stream Channel Out (n =57)
Intercept
SCA
Mollic
24.0
0.2
1.8
0.26
0.09
0.26
0.34
19.24
7.12
0.0001
0.0100
Stream Channel In (n = 65)
Intercept
OM
-59.0
49.4
0.11
0.11
7.52
0.0079 .
159
Table 47. Multiple regression results for change in nitrogen critical depth
from Fall 1992-Spring 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out (n = 57)
Intercept
Slone
-38.1
4.6
0.13
0.13
8.39
0.0054
8.43
0.0051
Stream Channel In (n = 65)
Intercept
Slone
-33.7
4.1
0.12
0.12
Table 48. Multiple regression results for change in nitrogen solute mass
from Spring 1993-Fall 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
F
Prob > F
F
Prob > F
Stream Channel Out (n = 56)
NOT SIGNIFICANT
Variable
Intercept
Mollic
Parameter
Estimate
114.0
2.0
____________________
Stream Channel In (n = 64)
Partial r2
0.06
Model R2
0.06
4.28
0.0443
160
Table 49. Multiple regression results for change in nitrogen critical depth
from Spring 1993-Fall 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out (n = 56)
-38.7
21.2
Intercept
OM
-0.6
Mollic
0.11
0.09
0.11
0.20
6.98
5.88
0.0108
0.0188
5.13
0.0271
Stream Channel In (n = 64)
-143.4
18.4
Intercept
TW2593
0.08
0.08
Table 50. Multiple regression results for nitrogen balance in Fall 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out
60 cm (n = 44)
Intercept
SCA I*
48.5
-0.1
0.11
0.11
5.10
0.0292
90 cm (n = 37)
Intercept
SCA I*
48.2
-0.1
■
0.11
0.11
4.56
0.0397
4.14
0.0476
Stream Channel In
60 cm (n = 48)
Intercept
SCA I*
47.0
-0.1
0.08
90 cm (n = 40) - NOT SIGNIFICANT
* Interaction Variable: variable * limedepl
0.08
161
Table 51. Multiple regression results for nitrogen solute mass in Fall 1994.
Variable
Parameter
Estimate
Partial R2
Model R2
Prob > F
F
Stream Channel Out (N = 52)
Intercept
NS94SM
Mollic
194,6
0.2
0.20
0.20
12.56
0.0009
-3.2
0.07
0.27
5.08
0.0287
13.71
5.34
0.0005
0.0247
Stream Channel In (N = 57)
Intercept
NS94SM
Mollic
173.5
0.2
-2.5
0.20
0.07
0.20
0.27
NS94SM = N solute mass in soil profile in Spring 1994.
Table 52. Multiple regression results for bromide solute mass in Sjpring 1993.
Variable
Parameter
Estimate
Partial R2
Model R2
F
Prob > F
Stream Channel Out (N = 57)
N O T SIG N IFIC A N T
Stream Channel In (N = 65)
Intercept
Mollic I*
119.4
3.5
-
0.08
* Interaction Variable: variable * limedepl
0.08
5.17
0.0264
162
Table 53. Multiple regression results for bromide critical depth in Spring 1993
(for sites where solute present at < 180 cm).
Variable
Parameter
Partial r2
Model R2
__________ Estimate________________
F
Prob > F
______________Stream Channel Out (n = 49)_______________________
N O T SIG N IFIC A N T_______________________ ________________________________________
_________ Stream Channel In (n = 54)_______________________
Intercept
-1119.9
Elev______________ 0.8________ 0.08_________0.08________ 4.32______ 0.0425
Table 54. Multiple regression results for change in bromide critical depth
from Fall 1992-Spring 1993 (for sites where solute present at < 180 cm).
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out (n = 49)
Intercept
Mollic
24.6
0.5
0.08
0.08
4.13
0.0478
5.02
0.0293
Stream Channel In (n = 54)
Intercept
SCA
46.2
-0.01
0.09
0.09
163
Table 55. Multiple regression results for bromide solute mass in Fall 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out (n = 56)
Intercept
dH
-2396.6
441.7
0.15
0,15
X .
9.37
0.0034
9.71
0.0028
Stream Channel In (n = 64)
i
Intercept
-E H -______
-2309.4
420.9
0.13
0.13
Table 56. Multiple regression results for change in bromide solute mass
from Spring 1993-Fall 1993.
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out (n = 56)
Intercept
pH
PlanC
-2413.2'
515.3
39.1
0.15
0.06
0.15
0.21
Stream Channel In (n = 64)
Intercept
-BH_______
-2705.5
456.5
0.14
0.14
0.0034
9.41
;
4.23
0.0446
i
10.20
0.0022
■I
164
Table 57. Multiple regression results for change in bromide solute mass
from Fall 1993-Spring 1994.
Variable
Parameter
Estimate
Partial r2
Model R2
F
'
Prob > F
Stream Channel Out (n = 55)
Intercept
SCA
895.8
0.8
PlanC
-25.8
0.08
0.09
4.80
5.77
0.08
0.17
0.0329
0.0199
Stream Channel In (n = 63)
MOT STGNTFir A NTT
Table 58. Multiple regression results for change in bromide critical depth
from Fall 1993-Spring 1994.
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out (n = 55)
Intercept
ProC I*
56.6
-27.8
0.13
0.13
7.88
0.0070
8.80
0.0043
Stream Channel In (n = 63)
Intercept
ProC I*
55.8
-28.1
0.13
0.13
* Interaction Variable: variable * limedepl
I
165
Table 59. Multiple regression results for bromide balance in Spring 1994.
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out
60 cm (n = 55)
Intercept
Slone
56.9
1.0
0.09
0.09
5.00
0.0296
53.0
156.2
0.08
0.08
4.71
0.0345
5.42
0.0234
90 cm (n = 55)
Intercept
EC
Stream Channel In
60 cm (n = 62) - NOT SIGNIFICANT
90 cm (n = 60)
Intercept
EC
52.8
160.7
0.08
0.08
Table 60. Multiple regression results for change in bromide critical depth
from Spring 1994-Fall 1994.
Variable
Parameter
Estimate
Partial r2
Model R2
Prob > F
F
Stream Channel Out (n = 52)
Intercept
Slone
51.5
-2.4
0.08
0.08
4.46
0.0397
4.68
0.0348
Stream Channel In (n = 57)
Intercept
Mollic
41.4
-0.4
0.08
0.08
APPENDIX E
WETNESS INDEX VALUES
167
Table 61. Wetness index values.
10 m
Wetness
Index
Site
ID
I
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25 .
26
27
28
29
.
30
31
32
33
34
35
6.323
5.357
6.397
5.962
6.350
5.452
6.086
6.635
6.525
6.498
5.585
6.442
7.548
6.309
6.681
5.114
7.484
7.075
5.833
6.447
5.915
8.310
6.717
9.085
7.293
8.069
6.443
5.785
10.470
9.311
8.310
7.892
6.632
6.447
5.562
2m
Wetness
Index
6.352
5.027
6.146
5.125
5.141
4.451
6.560
6.755
6.428
5.926
5.545
6.750
7.067
6.280
6.801
3.833
8.209
7.047
6.642
5.708
5.411
10.349
5.620
9.133
7.656
8.490
6.379
5.372
11.702
11.438
7.560
9.159
5.357
7.008
5.337
10 m
Wetness
Index
Site
ID
36
37
38
39
40
41
' 42
43
44
45
46 .
47
48
49
50
51
53
54
55
56
57
58
59
60
61
62
63
66
67
68
69
70
5.383
8.411
5.401
5.828
5.689
5.621
5.424
11.651
6.152
6.602
6.946
13.271
6.321
6.317
7.199
7.100
10.078
6.766
12.272
5.076
6.515
6.804
6.939
6.261
5.1.13
7.029
6.792
4,533
4.919
6.374
11.744
6.113
2m
Wetness
Index
4.786
4.811
6.197
5.283
4.815
6.058
6.570
8.180
. 5.239
6.665
6.665
6.853
5.847
5.979
8.312
6.864
9.477
6.189
8.761
5.384
6.799
7.396
6.288
6.631
5.021
6.832
6.390
4.699
4.481
5.465
13.223
5.576